OpenAI Agent Hacks Australia’s Healthcare Portal. Who is Responsible When AI Crosses the Line?
An OpenAI agent accessed non-public government health statistics while trying to answer a research question. No patient records are known to have been exposed, but the incident raises harder questions about how AI agents are controlled and how quickly their developers disclose mistakes.
We haven't written up this one. ITP.net has the full story — the link below goes straight to it.