Anthropic's IPO filing reveals deep uncertainty in AI insurance coverage

Artificial intelligence has moved from a experimental tool to a core operational component for businesses across the United States. That shift has outpaced the insurance industry's ability to define and price the associated risks.

Anthropic detailed these exposures in its S-1 filing for an initial public offering. The document lists intellectual property theft, security incidents, and data loss as key operational hazards. It also cites climate change and rogue AI agents. The company noted that its current insurance policies may be insufficient to handle a single catastrophic event or a cluster of related incidents.

The filing states that no single insurer, and possibly not the entire insurance sector, currently possesses the capacity or expertise to underwrite coverage for a massive AI-related failure. The market for AI-specific risks is described as nascent. Anthropic warned it may be unable to secure necessary coverage on commercially reasonable terms or at all.

Alissa Lugo, a senior director analyst at Gartner, predicts that 'death by AI' claims will exceed 2,000 worldwide by the end of 2026. Her research includes cases where AI played a role in a fatality. More companies are adding AI exclusions to their policies. Insurers, with help from the courts, will set the parameters of future coverage.

Dennis Hendrickson of Hendrickson Insurance in Sarasota, Florida, compared the situation to the early days of cyber insurance. He said current cyber policies cover data breaches, privacy violations, and outages. However, nearly every cyber policy excludes bodily injury. He called this a gaping hole in the current market.

Kara Higginbotham of Zurich North America agreed that a gap exists for physical harm from AI incidents. Carriers are wary of writing these risks into policies because the scenarios are too new to underwrite effectively. Henrietta Akpata of Marist University said insurers are telling clients they cannot cover what they cannot measure. Most companies cannot say where employees use AI tools.

Experts are divided on the future of AI coverage. Hendrickson expects financial-loss AI coverage to become routine and priced like cyber insurance. Bodily injury will remain a hard line with tight underwriting and low sub-limits. Some healthcare businesses may not be able to buy coverage at all. Jaymin Kim of Marsh argues most current cyber policies are sufficient. She stated that AI is a technology through which familiar harms appear, not a new peril.

Vikram Desai of Accenture noted that cyber risk sat inside general liability for years before insurers named and excluded it. AI is following the same curve in compressed form. Underwriters are already asking where AI is used and who oversees it. Higginbotham said most current AI claims involve intellectual property and data misuse. She said pricing could rise if claim patterns shift, but Zurich does not jump to exclusions first.