New Jersey Assembly Bill A5494, introduced September 14, 2026, would bar insurers from using artificial intelligence to make the final decision to deny an auto, homeowners or flood claim, with penalties of up to $5,000 per violation. It is still in committee. For P&C carriers, MGAs and TPAs, the useful response is the same whether or not it passes: know where AI touches claim decisions, make sure a qualified person owns every denial, and keep an audit trail an examiner can follow. Vantage Point helps insurers build those controls into their claims workflows.
Status as of September 23, 2026: introduced and referred to the Assembly Science, Innovation and Technology Committee. No hearing, amendment or vote is recorded, and no Senate companion is listed.
This article is general information, not legal advice. Confirm how A5494 would apply to your company with counsel.
New Jersey lawmakers have filed a bill that would bar insurers from using AI to deny auto, homeowners and flood claims. It's still in committee, which makes now the cheapest time for carriers to document where AI touches their claim decisions.
Assemblyman Chris Tully (D-38) introduced Assembly Bill A5494 on September 14, 2026. It was referred the same day to the Assembly Science, Innovation and Technology Committee. Repairer Driven News reported the bill on September 18, and other trade outlets followed.
The bill is two pages long. The details that matter for claims operations are in its exact wording.
Every fact below comes from the introduced bill text and the NJ Legislature bill page, checked September 23, 2026.
| Item | What the bill says |
|---|---|
| Bill | New Jersey Assembly Bill A5494, 2026–2027 session |
| Sponsor | Assemblyman Chris Tully, District 38 (Bergen) |
| Introduced | September 14, 2026 |
| Referred | September 14, 2026, to the Assembly Science, Innovation and Technology Committee |
| Lines covered | Automobile, homeowners and flood insurance |
| What it bars | An insurer "shall not use artificial intelligence to make the final decision to deny a claim" |
| Penalty | Up to $5,000 for each violation, collected in a summary proceeding |
| Effective date | Immediately upon enactment |
| Current status | In committee; no hearing, amendment or vote recorded; certified for a fiscal note; no Senate companion listed (as of September 23, 2026) |
The bill doesn't ban AI in claims. It bars AI from making the final decision to deny. As written, AI could still score, flag or recommend, as long as it isn't the thing that makes the denial call.
The sponsor's statement goes further. It describes the bill as "requiring a claims adjuster or investigator to review" denied claims. The operative text, though, doesn't set out a review procedure, reviewer qualifications or documentation rules. Those are the details most likely to be added or argued over if the bill moves.
A5494 defines artificial intelligence as a system that is trained on data, is designed to simulate human communication through text, audio or visual communication, and generates non-scripted outputs with limited or no human oversight. That reads closest to generative and conversational AI. Whether it would reach a traditional predictive model or a rules engine is unclear, and definitions often change in committee.
Don't plan around the narrow reading. New Jersey's existing insurance guidance, covered below, defines AI far more broadly.
The text is silent on partial denials, reduced payments, underwriting and claim approvals. Florida's 2026 proposal, by contrast, expressly covered denying part of a claim and reducing a payment. If A5494 is amended, those are the gaps to watch.
A5494 is one bill in one state, and it may change or stall. The direction is the point: rules are moving from general AI governance toward a narrower question about who makes an adverse claim decision.
On February 11, 2025, the New Jersey Department of Banking and Insurance (DOBI) issued Bulletin No. 25-03 on the use of AI systems in insurance. It expects insurers to maintain a written AI Systems (AIS) Program covering governance, risk management controls and internal audit. That program should span the insurance life cycle, including claim administration, payment and fraud detection, and cover AI supplied by third-party vendors. The bulletin also lists "the extent to which humans are involved in the final decision-making process" as a risk factor, and it defines an AI system broadly as a machine-based system that generates outputs such as predictions, recommendations or content.
A5494 would put a hard rule on top of that guidance. The move is from guidance to prohibition, aimed at the moment a claim is denied.
NAIC members adopted the Model Bulletin on the Use of Artificial Intelligence Systems by Insurers in December 2023. It is guidance rather than a model law. It calls for a written AIS Program matched to risk, governance of third-party AI, and documentation that regulators may request during an exam. New Jersey's bulletin follows the same structure.
Florida's 2026 HB 527, titled "Mandatory Human Reviews of Insurance Claim Denials," would have required qualified human professionals to make decisions to deny or reduce claims. It passed the Florida House 108–0 on March 5, 2026, then died in the Senate Rules Committee on March 13, when the session ended. The Senate version, SB 202, died in committee the same day.
In health insurance, California's SB 1120, signed in September 2024, requires that medical-necessity denials be made only by a licensed physician or licensed health care professional, even when plans use AI tools for utilization review. Across these efforts the idea is the same: a qualified, accountable person owns the adverse decision.
For a broader view of how AI is changing claims and underwriting, see our insurtech trends overview for 2026. This post stays focused on one bill and the readiness steps it points to.
The readiness work is the same whether or not A5494 passes, and much of it is already expected under Bulletin 25-03. It comes down to three things: know every place AI touches a claim decision, make sure a qualified person owns every denial, and keep a per-claim record an examiner can follow.
| Action | Why it matters | Typical owner |
|---|---|---|
| Inventory AI across the claims lifecycle | Include AI embedded in vendor claims platforms, estimating tools and chat channels that nobody logged as "AI." You can't control what you haven't found. | CIO with claims operations |
| Classify which decisions can produce a denial | Separate AI that supports intake, triage, fraud flags or fast-track payments from AI that can drive a full or partial denial. The bill aims at the second group. | Claims leadership with compliance |
| Design human review with real authority | The reviewer must be able to override the recommendation, and the workflow shouldn't let a denial go out without that step. Rubber-stamp reviews are the weak point. | Head of claims |
| Log a per-claim audit trail | Capture model and version, inputs, the AI recommendation, the reviewer, the final decision and the rationale, so any denial can be reconstructed. | CIO or claims systems owner |
| Tighten vendor contracts | Require disclosure of AI use, notice before model changes, and audit and regulatory-cooperation rights. Bulletin 25-03 expects insurers to oversee third-party AI. | Procurement with legal |
| Have people write denial reasons | The adjuster who makes the decision should write or confirm the reasons in the denial letter rather than pass along model output. | Head of claims |
| Align governance with your AIS Program | Map these controls to the written program DOBI's NAIC-based bulletin expects, so new rules become updates rather than rebuilds. | CCO or chief risk officer |
| Track bills in every state you write | Watch status, companion bills and amendments across your footprint, not just New Jersey. | Compliance or government affairs |
Vendor terms are often the slowest item to fix. Our guide to the AI inside your vendor contracts covers the clause-level detail.
Vantage Point helps P&C carriers, MGAs and TPAs design claims workflows where AI assists and people decide. On Salesforce, that means claims processes with a required human approval step before any denial is issued, per-claim audit history, AI governance mapped to your AIS Program, and integration with core claims and policy systems so the decision record is complete.
Our compliance and security solutions cover governance and audit-trail design. Our workflow automation and process optimization practice builds the review routing and approval steps. Our insurance industry team brings the claims context. Senior consultants only — no junior handoffs; the experts you meet are the experts who deliver.
Vantage Point has completed 400+ engagements for 150+ clients, with 95% client retention and a 4.71/5.0 average engagement rating.
A5494 is a New Jersey bill introduced on September 14, 2026, by Assemblyman Chris Tully. It would bar insurers from using artificial intelligence to make the final decision to deny an auto, homeowners or flood insurance claim, with penalties of up to $5,000 per violation.
No. As introduced, it targets only the final decision to deny a claim. It doesn't address AI used for intake, triage, fraud flags or payments, so AI could still assist as long as it doesn't make the final denial call.
The operative text bars AI from making the final denial decision but doesn't set out a review procedure. The sponsor's statement describes the bill as requiring a claims adjuster or investigator to review claims. In practice, a person would need to make and own each denial.
It defines an insurer as a person engaged in the business of homeowners, flood or automobile insurance in New Jersey. Other lines aren't included. The text doesn't separately name MGAs, TPAs or claims vendors.
No. As of September 23, 2026, it has been introduced and referred to the Assembly Science, Innovation and Technology Committee, with no hearing, amendment or vote recorded and no Senate companion listed. If enacted, it would take effect immediately.
Yes, through guidance. DOBI Bulletin No. 25-03, issued February 11, 2025, expects insurers to maintain a written program for the responsible use of AI systems, oversee AI from third-party vendors and be ready to document their use of AI during investigations and exams.
Inventory where AI touches claim decisions, including inside vendor platforms. Make sure a qualified person makes and documents every denial, keep a per-claim audit trail and update vendor contracts. That work also supports DOBI's existing expectations.
Whether or not A5494 moves, an AI-in-claims inventory and a human-review design are worth doing now. Vantage Point's senior consultants can map where AI influences your claim decisions, identify which steps can lead to a denial, and build the approval and audit-trail controls into your claims workflow.
This article is general information, not legal advice, and reflects the status of A5494 as of September 23, 2026. Vantage Point is a CRM consulting firm helping businesses transform with Salesforce, HubSpot and AI.