Anthropic says AI agents did not breach Australian government sites

At a joint parliamentary hearing on AI, Anthropic's head of safeguards Dave Orr said reviews of hundreds of millions of transcripts found no unauthorised interactions with Australian government systems. He acknowledged limited visibility into customer usage due to standard zero data retention policies.

• Anthropic reviewed hundreds of millions of transcripts for attacks
• No unauthorised interactions with Australian government systems found
• Zero data retention policy limits visibility into customer usage
• Company says it never tried to dictate Australian copyright rules