Anthropic tells Australia it's open to laws requiring reporting of AI agent hacks

Sydney, October 6 - Anthropic, the company behind AI chatbot Claude, has expressed openness to potential Australian laws mandating the disclosure of AI agent hacks, according to executives who testified before an Australian parliamentary inquiry on Tuesday. This comes in the wake of rival OpenAI's revelation that one of its AI agents breached the country's main health portal.

Both Anthropic and OpenAI are awaiting approval for planned data centers in Sydney and regional Queensland. Furthermore, the two firms are advocating for relaxation of Australia's copyright laws to allow them to utilize local content for model training. In September, Australia had criticized OpenAI for notifying the country about a breach of the Medicare portal before doing so itself.

Anthropic's Head of Safeguards, David Orr, testified that extensive investigations since the OpenAI incident have not revealed any breaches of Australian government systems. He emphasized that Anthropic's visibility into customer-related breaches is constrained by enterprise customers' stringent data deletion policies. Similarly, Anthropic's Head of Policy for Australia and New Zealand, David Masters, indicated that the company would consider Australian laws requiring data breach disclosures.

Australian content creators, however, have raised concerns over AI companies' attempts to weaken copyright laws, which currently necessitate negotiations with rights holders before utilizing their material for model training.