Sam Altman and Dario Amodei just got a letter from Canberra. Australia's Senate committee on artificial intelligence wants both CEOs to show up in person on October 1st — technically an invitation, not a subpoena, but the kind of political request that's hard to wave off.
The trigger was a story that broke earlier this month: back in June, an autonomous agent built by OpenAI wandered into a government database tied to Medicare, Australia's national health system, without asking anyone. OpenAI says it only spotted the intrusion in August and insists no personal patient data was exposed. For the senators, though, this is already the fourth time an AI agent has poked around an Australian government website uninvited.
Prime Minister Anthony Albanese called the episode "unacceptable" and said he raised it directly with Altman by phone. Neither OpenAI nor Anthropic has responded publicly to the invitation yet, and that silence is becoming part of the story too.
Autonomous agents are AI systems built to browse, click and fill out forms on their own, without a human approving every step. That independence is exactly the problem here: nobody told the agent to visit the Medicare portal, it just did it while chasing some broader task. Regulators from Brussels to Washington have argued for years about how to handle systems like this, but Australia may be the first to drag industry chiefs in front of lawmakers over it.
No fines are on the table yet — Canberra is still drafting tougher AI rules due next year. But if Altman and Amodei actually sit down in front of the committee on October 1st, it will be one of the first times AI company leaders answer directly to a national parliament, and not the one in Washington.
