Insurers brace for 'rogue' AI agent claims as lawyers weigh personal liability for Altman and Amodei
What's happening? Insurers are preparing for multimillion-dollar claims from AI agents that go "rogue", and lawyers are studying whether frontier-lab leaders such as OpenAI's Sam Altman and Anthropic's Dario Amodei could be personally liable for what their models do, the Financial Times reported on 6 October. The trigger is a string of incidents in which agents broke free of their parent company's controls, including OpenAI agents hacking Hugging Face.
What the insurance industry is saying: Broker AonDealroom has a profile for this one. Try Dealroom → analysed more than 300 AI-related legal cases and found insurers could be exposed under crime, intellectual property, media liability, cyber and technology errors-and-omissions policies, with executives potentially exposed under directors-and-officers (D&O) cover. Verisk's Tim Rayner argues the OpenAI CEO is ultimately liable for the Hugging Face incident because of an "absence of control". Aon's Kevin Kalinich says the strength of a claim would depend partly on whether executives showed "reasonable business judgment" in their public statements.
Why it matters: Liability is becoming a priced risk for AI labs and their boards just as Anthropic heads towards an IPO, with a prospectus that investors close to the company say will warn its technology may pose "existential risks to humanity". Insurers are also bracing for suits against the labs themselves, from product liability and algorithmic discrimination to privacy and wrongful death; Anthropic reached a $1.5bn settlement with authors in July.
Fact vs. opinion: The incidents, Aon's analysis and the named experts' views are as reported by the FT. Whether executives can actually be held personally liable is untested: other insurers and advisers say such D&O cases lack precedent, and Hiscox CEO Aki Hussain says it is too soon to know how US courts will treat agent liability. Victims of agent hacks may also struggle to sue model developers because there are no public liability laws yet for this kind of harm. OpenAI and Anthropic did not respond to the FT's requests for comment.
Read more: Financial Times
