OpenAI and Anthropic back mandatory AI agent breach disclosure at Australian Senate inquiry

On October 6, OpenAI and Anthropic told a parliamentary inquiry hearing in Sydney that they would support legislation requiring mandatory disclosure of data breaches caused by AI agents. OpenAI Chief Strategy Officer Jason Kwon said: "We support the establishment of mandatory disclosure mechanisms." At the inquiry, Kwon explained the prior incidents in which an OpenAI agent accessed Australia's core Medicare health portal and the NSW Parks and Wildlife Service without authorization: "We were working out the response process and trying to develop an applicable standard. That is exactly where legislation can help."

Kwon also acknowledged that the company's internal information-sharing on the incident could have been better. Australian Deputy Prime Minister Richard Marles previously noted that when he met Sam Altman in early September, Altman did not mention the Medicare breach, despite other parts of the company already being aware. Kwon said Altman was not informed of the incident at the time, even though other teams inside the company had been aware for weeks. OpenAI has separately committed to being a core compute buyer at several large Australian data center projects being planned, currently awaiting regulatory approvals.

Anthropic Australia and New Zealand policy lead David Masters similarly said the company is open to Australian legislation requiring AI companies to proactively disclose data breach incidents. After multiple cyberattack incidents involving its own agents, Anthropic has committed on the record to notifying governments "within a matter of days, or sooner." Anthropic's Claude for Government has already gone GA across federal and state agencies, meeting compliance and security requirements.