Your Car Accident Claim Might Now Involve an AI Fight You Didn't Expect

If you’ve filed a car accident claim recently, there’s a good chance artificial intelligence touched it before a human adjuster ever did — and a growing body of law is starting to address what happens when that AI gets it wrong.

How AI already shapes your claim, on both sides

On the insurer’s side, carriers increasingly use AI and machine-learning tools to triage claims, estimate vehicle damage from photos, flag claims as “high risk” for further scrutiny, and in some cases generate settlement offers with minimal human review. State insurance regulators have taken notice: as of 2026, at least eleven states have formally adopted the National Association of Insurance Commissioners’ Model Bulletin on the Use of Artificial Intelligence Systems by Insurers, which requires insurers to have a written AI governance program and be able to explain and audit AI-driven decisions that affect consumers (McDermott Will & Emery, 2026; NAIC Model Bulletin, official text, PDF).

That regulatory attention exists because courts are already testing it in litigation. In at least one bad-faith case, a court allowed discovery into an insurer’s internal AI claims-denial system — meaning the insurer had to turn over information about how the algorithm was built, trained, and used to evaluate real claims (Hunton Andrews Kurth, Insurance Recovery Blog). That matters because most states’ unfair claims settlement practices laws — modeled on the NAIC’s Unfair Claims Settlement Practices Act — already require insurers to conduct a reasonable, good-faith investigation before denying or lowballing a claim (NAIC Unfair Claims Settlement Practices Act, model text, PDF). An algorithm that denies claims in bulk with no meaningful human review is a natural fit for exactly the kind of bad-faith claim those statutes were written for.

On your side of the fight: the data your car is already collecting

The flip side is that claimants and their attorneys now routinely pull data the vehicles themselves generated. Nearly every car built after 2012 contains an event data recorder (EDR) — commonly called a “black box” — that captures pre-crash speed, braking, throttle position, seatbelt use, and airbag deployment timing in the seconds before and during a collision. These devices are federally regulated under 49 C.F.R. Part 563, administered by the National Highway Traffic Safety Administration (NHTSA, Event Data Recorders; 49 C.F.R. Part 563, eCFR). That data, combined with insurer telematics and app-based driving scores, is increasingly the deciding factor in disputed liability cases — and it can just as easily undercut an insurer’s AI-generated liability assessment as support it.

What this means practically

If your claim gets an unusually fast lowball offer, or a denial that doesn’t clearly explain its reasoning, that’s worth a second look — not necessarily because AI was involved, but because an unexplained, formulaic-feeling response is precisely the pattern regulators and courts are now scrutinizing. Preserving your vehicle’s EDR data promptly (before it’s overwritten or the car is repaired/totaled) is one of the most concrete steps a claimant can take early on.

This post is for general informational purposes and is not legal advice. Insurance regulation in this area varies significantly by state — consult a licensed attorney about your specific claim.

Sources