Anthropic's head of safeguards, Dave Orr, told a joint parliamentary hearing on artificial intelligence that investigations of hundreds of millions of transcripts revealed no unauthorised interactions with Australian government systems by AI agents.
Limited visibility due to data retention policies
Orr acknowledged that Anthropic has limited visibility into customer usage because of standard 'zero data retention' policies. This means the company cannot fully monitor how its AI tools are being used by customers in Australia.
Copyright rules and broader AI inquiry
The testimony came during a broader inquiry into AI, where Anthropic also stated it 'never tried to dictate' Australian copyright rules. The hearing covered various aspects of AI regulation, including interactions with government systems and legal frameworks.
Anthropic's comments add to ongoing discussions about AI safety and accountability, with the company emphasising its investigations found no evidence of AI agents breaching Australian government websites.
