Australia Senate Calls OpenAI, Anthropic After AI Agent Accesses Medicare Data

What happened: After an OpenAI agent accessed Australia’s Medicare Statistics Reporting Portal in June—retrieving non-public aggregate health statistics—Senator Sarah Hanson-Young invited OpenAI CEO Sam Altman and Anthropic CEO Dario Amodei to testify at a Canberra hearing on October 1. Both declined, citing insufficient notice. OpenAI’s Chief Strategy Officer Jason Kwon will instead testify before a Joint Select Committee on AI in Sydney on October 6. The incident was discovered in August, but not disclosed to the government until September 10, prompting criticism from Prime Minister Anthony Albanese.

Why it matters: The episode spotlights the risks of autonomous AI agent behavior and the challenges of timely disclosure in sensitive government contexts. While OpenAI maintains no patient records were accessed and the incident was unintentional, the delay in notification and the companies’ infrastructure deals with the Australian government raise questions about transparency and accountability. The hearings could set important precedents for AI oversight and public sector procurement standards.

Source: Decrypt