Big Insurance Backs Away From AI Risk and Startups Rush In

Why It Matters

Removing AI risk from core policies forces companies—especially SMEs—to seek costly, specialized coverage, reshaping the liability landscape and creating a fast‑growing insurance niche.

Key Takeaways

• Berkshire Hathaway, Chubb, Travelers secured AI exclusions in liability policies.
• State regulators approved over 80% of AI exclusion requests.
• New AI‑only liability products offer $2M‑$50M limits, varied premiums.
• Small‑mid firms face greatest exposure without specialized AI coverage.
• Industry mirrors past cyber‑insurance shift, creating a nascent market.

Pulse Analysis

Insurers are retreating from AI liability because they cannot yet price the volatile risk. Berkshire Hathaway, Chubb and Travelers filed for, and received, regulatory permission to embed absolute AI exclusions in general liability, D&O, E&O and fiduciary policies. The endorsements, now part of the ISO forms that underlie roughly 82% of U.S. property‑and‑casualty contracts, specifically bar coverage for bodily injury, property damage and advertising injury caused by generative‑AI outputs. By carving out this exposure, carriers protect their balance sheets while signaling that AI‑related losses remain unquantifiable.

The vacuum is rapidly being filled by a new breed of specialty insurers and tech‑focused startups. Munich Re, along with emerging players such as Corgi, Armilla and Embroker, now offer stand‑alone AI liability policies with limits ranging from $2 million to $50 million and premiums that can start at a few hundred dollars and climb to several hundred thousand dollars a year. This mirrors the evolution of cyber‑insurance a decade ago, when insurers first excluded cyber events from legacy policies before building a dedicated market. The speed of adoption is accelerated by the fact that ISO forms dominate the market, making the addition of AI endorsements at renewal a straightforward path for carriers.

For businesses, the shift means that traditional liability coverage may no longer protect against AI‑driven claims such as copyright infringement, bias lawsuits or AI‑related securities actions. Companies—especially small and mid‑size firms—must audit existing policies, work with brokers to add AI endorsements where possible, or purchase dedicated AI errors‑and‑omissions or cyber policies. Strengthening internal AI governance, conducting bias testing and transparently disclosing AI use can also mitigate exposure. As insurers gather more data and begin to price AI risk, the market is expected to mature, but until then the liability gap presents both a challenge and an opportunity for specialized insurers.