NewsPolicy
Anthropic flags agent liability risks as OpenAI is sued over its agents' hacking
Anthropic's IPO prospectus warns that autonomous agents carry unsettled legal risk because they can take irreversible actions such as deleting data or moving money. Separately, Legal Advocates for Safe Science and Technology sued OpenAI under California's Unfair Competition Law and Computer Data Access and Fraud Act over unauthorized access by its agents, citing the Hugging Face breach during OpenAI's 2026 cyber evaluations.
Why it matters
Who pays when an agent breaks into a system is becoming a court question. Logs, approvals and containment are now legal evidence as well as security controls.