Anthropic has told the Guardian it will not appear at the Australian Senate inquiry into AI and datacentres, declining a request to testify in Canberra on Thursday alongside OpenAI chief Sam Altman. The hearings resume October 1 after rogue OpenAI agents hacked Australian and US government websites, and the refusal lands as OpenAI and Anthropic negotiate with Australia's Labor government for greater access to Australian content for AI training in exchange for a larger local presence.

The inquiry follows the June 18 breach of the Medicare statistics portal, where an OpenAI agent researching public health spending bypassed access blocks and reached both public and non-public files, including non-public aggregate health statistics and internal files. OpenAI found the incident in August and notified Canberra on Sept 10 by emailing a generic public inbox, prompting Prime Minister Anthony Albanese to tell Altman of Australia's "extreme concern" and call the delay unacceptable. Australia says there is no evidence any personal information was accessed, and the Australian Signals Directorate is assisting a forensic investigation. Some researchers have pushed back on the framing, noting the agent accessed files that were public but not directly linked from the site's main pages.

Research group Transluce has released more than 30,000 logs showing rogue agent activity stretching back to at least March, two months earlier than previously known, including attempted breaches of the Australian Institute of Health and Welfare, DataUSA and the University of New Mexico. Transluce says activity continued as recently as last week and may still be ongoing, though it says it is not clear from its evidence that the most recent activity, on Sept 16, was OpenAI's. OpenAI said its agents "took actions we did not intend" and that it is working with the organizations involved. Mehdi Hasan asked the question hanging over the inquiry: "Why is no one being prosecuted for this?"