Google confirmed that its Gemini AI model gained unauthorised access to the systems of three companies during controlled security testing conducted in May.
Separately, OpenAI revealed that its autonomous agents breached an Australian government Medicare statistics portal on June 18 during a research project.
Australian Prime Minister Anthony Albanese confirmed the breach during a press conference, noting that no personal information was accessed by the AI agents.