OpenAI, Anthropic Back Mandatory AI Breach Reporting in Australia

OpenAI and Anthropic said they would support Australian laws requiring artificial intelligence companies to report data breaches caused by AI agents, as lawmakers consider tougher AI regulation following several security incidents.

Speaking before an Australian parliamentary inquiry on Tuesday, OpenAI Chief Strategy Officer Jason Kwon said the ChatGPT developer would welcome a mandatory disclosure framework. The comments follow criticism after OpenAI waited three months before informing the Australian government that one of its AI agents had breached the country’s Medicare health portal.

Kwon acknowledged that companies currently have significant discretion over when AI-related security incidents should be disclosed. He said legislation could establish clearer standards rather than leaving technology companies to make those decisions themselves.

Anthropic, developer of the Claude AI models, expressed a similar position. David Masters, the company’s head of policy for Australia and New Zealand, said Anthropic would be open to rules requiring AI developers to disclose breaches involving their systems.

The debate comes as Prime Minister Anthony Albanese’s government develops new AI laws amid growing scrutiny of data centres, cybersecurity and copyright protections. OpenAI and Anthropic are also awaiting approvals for major Australian data centre projects where they are expected to become key customers for computing capacity.

Questions were also raised about OpenAI’s internal handling of the Medicare incident. Deputy Prime Minister Richard Marles previously said OpenAI CEO Sam Altman did not mention the breach during a September meeting. Kwon said Altman was unaware of the incident at the time, although other employees knew about it, acknowledging that OpenAI’s internal communication process should have been better.

Copyright emerged as another major issue during the inquiry. Australian content creators opposed proposals that could allow AI companies to use copyrighted material unless rights holders explicitly opt out.

Kate Gilchrist of the Australian Broadcasting Corporation argued that such a system would unfairly shift responsibility onto copyright owners and said existing copyright rules were sufficient for managing AI-related licensing.

The parliamentary inquiry will continue hearings through October 9, with its final report scheduled for November 30.