Insurers Exclude AI Liability, New Market Emerges

PYMNTS reports that several major insurers have sought and obtained regulatory approval to exclude AI-related damages from standard corporate policies. PYMNTS cites The Information in saying state regulators approved more than 80% of those exclusion requests, with Florida, Connecticut and Maryland approving the highest numbers. PYMNTS reports that the Insurance Services Office (ISO) introduced two optional endorsements covering AI-related harms. PYMNTS says the endorsements address bodily injury, property damage and personal and advertising injury tied to generative AI outputs, including defamatory content, intellectual property infringement and physical damages traceable to AI errors. PYMNTS cites Policyholder Pulse in reporting that some carriers, including Berkley, have implemented absolute AI exclusions in lines such as directors-and-officers, errors-and-omissions and fiduciary liability. PYMNTS also reports that ISO forms underpin roughly 82% of U.S. property-and-casualty policies and that many carriers are expected to attach endorsements at renewal.

Industry-pattern observations: Risk transfer for AI-driven products depends on measurable failure modes, reproducible audit trails and legal attributions, not just model performance. Companies and underwriters often rely on logs, model provenance and incident reconstruction to assess causation. When standard policies exclude AI-caused harms, buyers seeking coverage typically require narrower, well-scoped contracts, stronger testing and clearer post-incident forensics. For practitioners, this increases the value of observability, model documentation and reproducible evaluation pipelines when negotiating third-party risk terms.

Reporting frames these exclusions as a near-term market response to uncertainty about liability stemming from generative AI outputs and model training practices. The move shifts a portion of downstream legal and financial exposure away from mainstream carriers, which may accelerate entrants-insurtechs and specialty underwriters-offering bespoke AI-liability products. For enterprises using or integrating generative models, publicly available coverage terms will become a factor in vendor selection and risk management.

For practitioners: monitor state-level regulatory filings and ISO adoption rates at renewal windows, disclosure language in standard policy forms, announcements from specialist insurers and any legislative or regulatory efforts that attempt to standardize AI liability definitions. Observers should also track whether market entrants tie premiums to concrete technical controls such as model provenance, adversarial testing and explainability metrics.