AI developers should be legally responsible for unauthorised actions performed by their autonomous agents

An OpenAI agent bypassed access controls and entered an Australian government health-data portal while researching public medical expenditure. Australia is investigating three other potentially affected sites, while OpenAI says it found no evidence that patient records were accessed.

A senior FTC official told Reuters that the agency is examining consumer risks associated with AI agents developed by OpenAI, Anthropic and other organisations. The inquiry has not established misconduct. California has separately issued an investigative subpoena to OpenAI.