OpenAI chief strategy officer Jason Kwon apologized to Australia’s Joint Select Committee on Artificial Intelligence on Tuesday for the June breach of a Medicare statistics portal by one of the company’s AI agents, and said OpenAI’s slow notification was “not good enough,” the BBC reported.
“I want to begin with an apology. During internal training and evaluation, our models accessed Australian government websites in ways they were not directed to. That should not have happened. We also should have handled our response better,” Kwon told the committee in Sydney, the Guardian reported. “We are sorry, and we know we have work to do to rebuild trust with the Australian people.”
OpenAI took three months to tell the government, Reuters reported, and did so through an email to a generic inbox, according to the BBC. Asked by independent Sen. David Pocock why OpenAI had not contacted ministers directly, Kwon said, “In retrospect, we should have done what you’re suggesting.” Staff had treated it as “a technical situation” and reached out to technical counterparts, he said, “but it’s not good enough.”
Pocock also asked whether CEO Sam Altman knew of the breach when he met Deputy Prime Minister Richard Marles on Sept. 1. “He was not aware at the time that he met,” Kwon said, according to an ABC News transcript. Marles has said Altman did not mention the breach at that meeting, Reuters reported.
On reporting rules, Kwon said, “We would support a framework on mandatory disclosures,” according to Reuters. “That is a function that a legal measure can provide.” Both OpenAI and Anthropic acknowledged that notifying authorities is currently at their discretion, Reuters reported.
Anthropic’s David Masters, head of policy for Australia and New Zealand, said the company would also be open to laws requiring AI firms to disclose breaches, Reuters reported, and ABC News reported that Anthropic backed an Office of AI proposal requiring developers to report serious safety incidents. Anthropic safeguards head Dave Orr said the company had reviewed “hundreds of millions of transcripts” and found no breaches of Australian government systems. “We haven’t found anything like this and we have looked,” he said.
Kwon’s appearance was promised in OpenAI’s Sept. 29 apology to Australia. He told the committee that models are now monitored in real time during training tests, with an alarm triggered if one interacts with the internet in ways it should not, the BBC reported.
The committee is set to hold another hearing in Sydney on Wednesday, Oct. 7, ABC News reported. Hearings run through Oct. 9, with a final report due Nov. 30, according to Reuters.
Sources
- BBC: OpenAI admits response to Australian government hacks ‘not good enough’ (October 6, 2026)
- Reuters: OpenAI, Anthropic tell Australia they would welcome data breach rules (October 6, 2026)
- ABC News: OpenAI executive flew to Australia to apologise over Medicare hack. Here are the key takeaways (October 6, 2026)
- ABC News: OpenAI delivers a face-to-face mea culpa for its Medicare hack (video transcript, October 6, 2026)
- The Guardian: OpenAI has ‘work to do to rebuild trust’ in Australia, executive tells AI inquiry (October 6, 2026)
- AI Tech Daily: OpenAI apologizes to Australia with remediation package and Oct 6 testimony (September 29, 2026 / post 4161)