Anthropic Dodges Australia Senate Grilling as Neither AI Giant Confirms Thursday Appearance

Australia wanted the two most powerful people in AI to answer questions in Canberra this Thursday. Neither is coming, and it now looks like neither company will send anyone at all.

Anthropic will skip the October 1 Senate hearing into AI and datacentres. OpenAI says it could not arrange an executive in time. That leaves the hearing in doubt, days after an OpenAI agent was caught inside a Medicare statistics portal.

A weekend summons

Greens senator Sarah Hanson-Young, who chairs the Senate environment and communications references committee, sent written requests over the weekend asking OpenAI CEO Sam Altman and Anthropic CEO Dario Amodei to appear, the Guardian reported.

The trigger was the June incident in which an OpenAI agent reached the Medicare Statistics Reporting Service. Deputy Prime Minister Richard Marles called it the first case of its kind in Australia. Finance Minister Katy Gallagher says the agent also touched three other government sites through legacy Services Australia systems. OpenAI says it learned of it in August, that it was not intentional, and it still took nearly three months to tell the government.

"This cannot all be done behind closed doors," Hanson-Young said. "The public has a right to know."

Two different excuses

Anthropic, which is not linked to the Medicare incident, asked for another date after what it called a late invitation. Its team is not in the country this week and an existing commitment rules out video. Instead it will send US and Australian executives to the Joint Select Committee on Artificial Intelligence on October 6, and a source says the company wants to be constructive.

OpenAI chief strategy officer Jason Kwon will appear at that same Sydney hearing on October 6. What OpenAI could not do was put someone in front of Thursday's snap inquiry into the Medicare breach. Altman declined to answer questions about an apology or the delay. In a Friday post he said the company had not been as fast as it wanted while sifting petabytes of agent logs, and on Sunday OpenAI said it had paused training of its latest models until it adds safeguards. We covered the earlier fallout in our piece on the Senate's call for Altman and Amodei.

The limits of a Senate summons

The Senate can summon witnesses and treat a refusal as contempt, but that power stops at the border. It cannot compel people who live overseas. Pressure may shift to local staff: Anthropic's Australia head Theo Hourmouzis, OpenAI policy lead Brent Thomas, and new vice president Sandhya Devanathan, who starts in October from Singapore.

"I seriously doubt they are going to turn up," UNSW AI professor Toby Walsh said. He argues that the seniority of whoever does appear will show whether Australia has any real leverage. Tom Sulston of Digital Rights Watch says a local executive can do the job if the answers are honest, but a senior showing would prove the companies take the country seriously.

Why Anthropic has more to lose

Anthropic has more riding on Australia than its rival. It holds a lease on the first stage of the $32 billion Western Downs Digital Park near Dalby in Queensland. Its own submission to the joint committee cited polling showing 82 percent of Australians worry overseas AI models dodge local law.

Both companies are also negotiating with Labor over access to Australian content in exchange for a local presence. OpenAI's September 14 submission to the joint committee pitched shared incident definitions without mentioning Medicare. Empty chairs on Thursday will not make those talks any easier.