Anthropic’s IPO filing indicates that existing insurance may not cover liabilities arising from the unintended behavior of its models, and that no insurer may have sufficient capacity to cover a catastrophic AI-related incident. Gartner estimates that more than 2,000 deaths in which AI played a role could be recorded worldwide by the end of 2026. Cyber insurance generally covers data breaches, service outages, deepfakes and financial fraud, but often excludes bodily injury or physical damage, leaving a significant gap. Insurers and experts are divided on how the market will develop: some believe AI will be incorporated into existing policies, while others anticipate a separate category with tighter underwriting, higher restrictions and human-review requirements. Companies that cannot clearly track their AI use could also face coverage exclusions.
Source: Read the original article
