Anthropic IPO Filing Acknowledges Some AI Risks May Be Uninsurable

Anthropic's IPO filing lays out a long list of risks, from intellectual property theft and data loss to rogue agents and even an existential threat to humanity. Woven through that language is a more basic question: can any policy cover the ways AI might go wrong? Even with more cash than nearly any company in America, Anthropic could not answer that with confidence.

In its S-1 filing, the company warned that its models may act in ways no user or developer intended, exposing it to substantial liability, and that its policies may not provide enough coverage-especially when a single incident or cluster of incidents causes disproportionate harm. At the scale of a catastrophic AI incident, the filing states, no single insurer-and possibly the industry as a whole-may have the capacity or expertise to underwrite coverage matching that exposure. The market for AI-specific risks is young and evolving, and the company may be unable to obtain coverage on commercially reasonable terms, or at all.

The concern extends well beyond chatbots booking flights or hunting for grill deals. Alissa Lugo, a senior director analyst in Gartner's legal and compliance practice, says her research shows "death by AI" claims-cases in which AI played some role in a fatality-could exceed 2,000 worldwide by the end of 2026, and the number is expected to grow. Dennis Hendrickson of Hendrickson Insurance, an independent brokerage in Sarasota, Florida, said cyber is the policy doing the most to cover AI today, but it responds mainly to data breaches, privacy violations, system outages and deepfake wire fraud. Nearly every cyber policy excludes bodily injury, he said, calling that a gaping hole.

Kara Higginbotham, head of professional liability and cyber for U.S. national accounts at Zurich North America, said every line of business is handling AI differently and a coverage gap generally remains for bodily or physical harm from an AI-related incident, because carriers must be able to underwrite the risk. Henrietta Akpata of Marist University said most companies cannot say where their employees use AI tools-the first thing to fix. Jaymin Kim of Marsh's cyber practice said the firm is not seeing many AI claims and argued AI is not a peril in itself but a technology through which familiar harms surface. Vikram Desai of Accenture compared AI to cyber risk, which sat inside general liability until insurers named, excluded and then sold it back-a curve he said AI is on, but compressed.

Source: CNBC