Australian inquiry chair says OpenAI must explain AI access to private data

The Joint Select Committee on Artificial Intelligence will hold four days of hearings, taking evidence from tech companies as well as unions, employer groups, banks and industry experts. Tuesday's hearing will include representatives from copyright and artists' groups, examining access to data used to train AI models, and the Australian Broadcasting Corporation, which warned in its submission of a "cannibalisation" of journalism and demanded AI companies be subject to the same copyright, defamation and privacy rules as news outlets.

OpenAI's chief strategy officer, Jason Kwon, will front the hearing alongside Adam Cohen, the company's head of economic policy, and Asia-Pacific national security lead, Peter Anstee. OpenAI last week apologised for the Services Australia incident and the manner in which it informed the government, saying it was "working to do better".

But Jo Briskey, the Labor MP chairing the committee, said the company still had questions to answer. "We have heard what happened, we understand what has occurred, and there's some pretty problematic issues around not only the fact that these AI agents did access non-public data, but the reality was also that they took far too long to to notify us," she said. "We've seen OpenAI make its public apology to Australians. That's important ... my focus is on what do they do next? How do they assure Australians that they that this won't happen again? And that's a difficult question."

Briskey said her starting point would be to interrogate how to balance the potential risks and rewards of AI, such as economic benefits as well as cyber security and defence applications. She said she would also raise concerns about online safety, scams, and how any benefits would be spread among all Australians. "I don't think we get the benefit unless we can assure and mitigate against the risks," she said.

Independent senator David Pocock, who is also on the committee, is demanding answers, saying OpenAI "still have a lot they need to answer" about its AI agent accessing Services Australia data on Medicare and the company's "appalling" tardiness in notifying the federal government. He plans to seek more answers from OpenAI about the incident.

"You've got a bunch of companies that are essentially self-regulating with Donald Trump. I don't think many people would trust Sam Altman and OpenAI ... This is supposedly a company the Australian government wants to welcome with open arms and do deals with," Pocock said. "The Medicare hack stuff will be pretty well looked at through the committee process. It's been a fairly appalling response from them."

Altman, in an interview with Politico published on Monday, said OpenAI "believe that the world should accept some bad things happening for the benefits of this technology and people having the agency."

Pocock has raised alarm about the potential impacts on artists, creatives and copyrighted material and said he was still to be convinced about the technology. "There's still big questions over the long-term benefit of all this for Australia ... Nobody has been able to answer that," he said. "[The government] is chasing this GDP sugar hit, but we haven't hit the big part of the datacentre buildout yet, and we're already [seeing] inflation going up. They want to raise GDP, but after the buildout phase, how do we get a return? Looking at how much tax big tech companies pay, it's not much, and I don't see how it'll change."

Pocock is keen to interrogate the government plans to address copyright issues. Anthropic, in its submission, suggested a form of "opt-out" model where rightsholders could request their data not be scraped; an idea SBS, in its submission, claimed was "routinely ignored or bypassed". Pocock said it was "hard to cop this industry saying it's too hard to deal with rightsholders when they're heading for multi-trillion-dollar valuations", and said the government should "force them [AI companies] to negotiate deals."

Editor's Summary

OpenAI faces an Australian parliamentary inquiry this week over an AI agent's access to non-public Services Australia data, including Medicare information, and over delays in notifying the government. The committee will also examine copyright, data access and AI regulation, with Anthropic, Microsoft, Google, media and artists' groups due to appear. OpenAI has apologised and said it is working to do better, while committee members say it must explain how it will prevent a recurrence.