States Approve Most AI Exclusion Requests, Spurring Standalone AI Insurance Market

TL;DR: State regulators have green‑lighted insurers’ requests to strip AI‑related damages from standard liability policies in more than four out of five cases, pushing businesses toward a nascent standalone AI insurance market. Coverage limits for those niche policies range from $2 million to $50 million, with premiums from a few hundred to several hundred thousand dollars a year.

Insurers have begun to remove AI‑related damages from standard general liability policies, saying they cannot yet price the risk. State regulators in Florida, Connecticut and Maryland approved the majority of requests, with over 80% of filings granted. This shift follows the introduction of two optional endorsements by the Insurance Services Office that let carriers add absolute AI exclusions.

The Insurance Services Office creates the policy templates that underlie about 82% of U.S. property and casualty insurance. When carriers attach the new AI exclusion endorsements to those forms, the change applies to a large share of the market. The exclusions took effect as early as January in several states.

State regulators approved over 80% of insurers’ requests to exclude AI-related damages from general liability policies. This figure comes from filings reviewed by state insurance departments. The high approval rate shows regulators are allowing carriers to retreat from covering AI risk.

ISO forms support roughly 82% of the U.S. property and casualty market, meaning the exclusions could reach a broad base of businesses. Because many policies are built on those templates, the change can be applied quickly at renewal. Carriers such as Berkley have already added the endorsements across directors and officers, errors and omissions and fiduciary liability lines.

Standalone AI liability policies now offer coverage limits between $2 million and $50 million. Annual premiums for those policies range from a few hundred dollars to several hundred thousand dollars, depending on limit and risk profile. The market mirrors the early cybersecurity insurance wave a decade ago.

Firms that rely only on standard liability coverage may find AI‑related claims uncovered, leaving them exposed to costs from defamation, IP infringement or bias allegations. Insurers say they lack sufficient loss data to price AI risk, so they are transferring the exposure to policyholders. Companies should review policy endorsements with brokers and consider technology errors and omissions, cyber liability or standalone AI policies.

Regulators may issue guidance on how broadly the AI exclusions apply, and courts will begin to interpret the language in liability disputes. The growth of standalone AI insurance will depend on whether premiums stabilize as loss data accumulates. Monitoring these developments will show whether a mature AI liability market emerges or if coverage gaps persist.