US Insurers Seek to Exclude AI Liabilities From Policies

Several major insurance companies are requesting permission from U.S. regulators to exclude liabilities arising from the use of artificial intelligence tools, such as chatbots and agentic AI, from their corporate policies. W. R. Berkley Corporation, along with Great American and Chubb, seeks to block claims involving the actual or alleged use of AI to avoid systemic risks where a single model failure triggers thousands of simultaneous claims.

Industry leaders describe the internal workings of AI as a "black box," citing the unpredictability of AI hallucinations and the threat of correlated losses. High-profile incidents contributing to this shift include a $110 million lawsuit against Google after its AI Overview falsely accused a solar company of legal trouble and a £20 million loss for engineering firm Arup due to a deepfake executive scam.

Response among carriers varies. While some firms like QBE and Chubb have introduced limited endorsements for specific issues, such as fines under the EU AI Act, Mosaic Insurance has declined to underwrite risks from large language models like ChatGPT entirely. American International Group Inc. filed for the option to implement exclusions and noted that generative AI is a wide-ranging technology likely to increase future claims, though the company stated it has no immediate plans to implement these exclusions at this time.