Australia's doctors want answers from OpenAI. AMA Victoria issued a media release on Oct. 6 demanding the company explain how one of its AI agents gained unauthorised access to a Services Australia Medicare statistics portal in June 2026 — and why authorities were not notified until September, nearly three months later.
The statement is timed to this week's appearance by OpenAI executives before Australia's Parliament Joint Select Committee on Artificial Intelligence, where the incident is expected to feature prominently.
The government has said no personal patient information was accessed. But AMA Victoria president Dr Simon Judkins warned the deeper question is what happens if an AI agent next reaches a system holding sensitive clinical data rather than aggregate statistics.
The association is pressing Victorian and federal governments for stronger safeguards around hospital records, My Health Record and e-prescribing against AI systems — and for faster intrusion detection and reporting.
The claims are, at this stage, the doctors' association's demands and characterisation of the incident, not an adjudicated finding of harm. But they land at a moment when regulators worldwide are asking the same question: who watches the agents?
