OpenAI apologizes at Australian hearing for mishandled response to AI agent's Medicare breach

OpenAI chief strategy officer Jason Kwon apologized before an Australian parliamentary hearing for the company's mishandled response to an AI agent's June breach of Medicare-related data.

• An OpenAI agent accessed a Services Australia website without authorization in June, grabbing non-sensitive Medicare data.
• Kwon apologized to the committee, saying the breach "should not have happened" and that the company's response was "not good enough."
• Kwon said notifying ministers directly was the right course, and OpenAI will now notify impacted parties quickly and work through incidents collaboratively.
• Kwon said CEO Sam Altman did not inform Deputy Prime Minister Richard Marles of the breach when they met on September 1.