Part 2 of 4 in a Field Guide series on AI and claim denials.
No state has enacted a law that explicitly bars P&C insurers from denying a claim on the basis of AI alone. What exists instead is a set of proposed bills and regulatory guidance. Beneath both sits adjuster licensing law, which predates AI and may already do much of the work a new statute would do.
Fl and NJ have proposed AI limits on P&C claims
Florida HB 527 and its Senate companion, SB 202, would have allowed insurers (and, in the House version, HMOs and workers' compensation carriers) to use AI to help process claims, but not as the sole basis for denying a claim or reducing a payment, and would have required that decision to be made by a "qualified human professional." That person would have had to analyze the claim independently of the AI system, review the accuracy of its output, and determine that the claim was not payable under the policy. Written denials would have had to give the reviewer's contact information and a unique identifier rather than the reviewer's name, and the insurer's records would have had to name that person. Both bills died in Senate committees in March 2026. The Legislature had considered a similar bill, CS/SB 794, in 2025.
The Florida bills did not invent a new kind of reviewer. They defined the qualified human professional as someone who already has authority under the Florida Insurance Code to adjust or deny the claim. In practice, that is a licensing question.
New Jersey's A5494 (2026–2027 session), introduced in September 2026, takes a narrower approach for homeowners, auto, and flood claims. The bill targets only the denial itself: AI could not make the final decision to deny. It has been referred to the Assembly Science, Innovation and Technology Committee.
Regulators have set governance expectations for AI
The NAIC's Model Bulletin on the use of AI systems by insurers, adopted in December 2023, asks insurers to maintain a written program governing their AI systems. Controls are meant to scale with the potential for harm to consumers, and insurers should expect regulators to ask about their governance and internal controls. The bulletin notes that insurers use AI across the insurance life cycle, including claim management and fraud detection. Many states have adopted it. It is a governance standard. It does not say who must make an individual claim decision. It does, however, treat the extent to which humans are involved in the final decision as a factor in how strong an insurer's controls should be.
Colorado has gone further for one P&C line. Under SB21-169, which covers claims management among other insurance practices, the Division of Insurance extended its algorithm governance and risk-management regulation to private passenger auto insurers effective October 15, 2025. The regulation is aimed at unfair discrimination and addresses how models are governed and tested. It does not address who decides a particular claim. However, to the extent a denial relies on a model that uses external consumer data, the statute may be triggered.
Adjuster licensing is a hidden limit on AI usage in Claims
Many states license some category of adjusters. According to the NAIC's State Licensing Handbook, 40 states license public adjusters, who represent policyholders, and 33 license independent adjusters, who handle claims for insurers on contract. Fifteen states require company adjusters to be licensed. Where a license is required, “adjusting” a claim is an act the law assigns to a licensed individual. This, of course, begs the question of exactly what it means to adjust a claim.
Texas has drawn the connection to AI directly. In a June 2026 bulletin on the use of AI, the Texas Department of Insurance reminded regulated entities that Chapter 4101 of the Insurance Code reserves certain acts to licensed adjusters, including investigating or adjusting losses and supervising the handling of claims. The same bulletin said that when AI is used to make a consequential decision, the department expects a person to review and agree with all decisions before action is taken. That is guidance rather than statute, but it rests on an existing licensing law.
What this means for claims organizations
Licensing may help explain why P&C has no AI-denial statute, but it leaves gaps. In most states it does not reach company adjusters. It also says who may adjust a claim, not how a licensed adjuster should use a model's output. An adjuster who accepts every AI recommendation without review may hold the required license and still fall short of the independent analysis the Florida bills described.
For claims leaders, the practical reading is modest. The Florida bills and the Texas bulletin describe the same arrangement. AI supports the file, and a person with authority makes the adverse decision and records the basis for it. Carriers that already handle denials this way should need few changes if a P&C statute passes. Carriers that rely on company adjusters in states that do not license them have more to confirm. So do carriers where AI output can move a claim toward denial without a documented review.
Sources: Florida HB 527 and SB 202 (2026), bill text, House staff analyses, and bill histories; Florida CS/SB 794 (2025). New Jersey A5494 (2026–2027 session). NAIC Model Bulletin: Use of Artificial Intelligence Systems by Insurers (adopted December 4, 2023), and the NAIC implementation map for the bulletin. Colorado SB21-169 (2021) and Colorado Division of Insurance Amended Regulation 10-1-1 (effective October 15, 2025). NAIC State Licensing Handbook, Chapter 18 (Adjusters). Tex. Ins. Code ch. 4101. Texas Department of Insurance Commissioner's Bulletin B-0003-26 (June 12, 2026).
This is Part 2 of 4. Part 3 covers what gets carriers sued over automated claims decisions.
