OpenAI and Anthropic have told an Australian parliamentary inquiry that they would support stronger rules requiring artificial intelligence companies to report serious data breaches.
Representatives from both companies appeared before the inquiry in Sydney on Tuesday as Australian lawmakers examined the risks posed by increasingly capable AI systems.
The hearing comes after an OpenAI AI agent accessed an Australian government website without authorisation. The incident involved a Medicare statistics portal and has raised concerns about how AI systems are monitored when they are allowed to interact with websites and online services.
OpenAI chief strategy officer Jason Kwon apologised during the hearing and acknowledged that the company should have informed Australian authorities sooner about the incident.
Kwon said OpenAI had made changes to its systems following the breach. The company has also said it wants to rebuild trust with Australians following the incident.
Anthropic, another major AI developer, told the committee that it would support mandatory reporting requirements for serious incidents involving AI agents. The company said its current approach to reporting such incidents was largely voluntary.
The comments could strengthen calls for Australia to introduce specific rules covering AI agents that can independently perform tasks online.
At present, companies have significant discretion over when and how they disclose incidents involving their AI systems. Mandatory reporting could give governments faster access to information about breaches and potentially allow authorities to respond before similar incidents spread.
The parliamentary inquiry is also examining wider issues surrounding AI, including copyright, safety, investment and the effect of the technology on Australian industries.
The hearing comes as Australia attempts to develop a regulatory framework for increasingly powerful AI systems while also attracting investment in data centres and technology infrastructure.
The inquiry is expected to continue examining the issue before making recommendations to the government.
For Australian consumers and businesses, the debate could eventually lead to clearer obligations for companies operating AI systems that can access sensitive information or interact directly with government and private-sector websites.
