Anthropic briefed critical infrastructure providers on AI threat — but didn’t discuss Mythos
During a whirlwind tour of Australia, the Claude developer warned AI could bring down critical services — but neglected to mention the model that has spooked regulators and markets.
During its high level visit to Australia earlier this month, Anthropic quietly briefed the organisations in charge of Australia’s most critical infrastructure on the threat of AI-powered cyberattacks but declined to mention the imminent arrival of its new model Mythos.
Capital Brief can reveal Anthropic met directly with national service providers last week to warn how bad actors could use cutting-edge AI models to disrupt the networks underpinning critical services like energy, water, telecommunications, health, and transport in Australia, urging them to prepare for a tougher cybersecurity environment.
The Department of Home Affairs confirmed the meetings and said it was currently preparing its own supplementary briefings for infrastructure owners and operators as it looks to fend off sophisticated future attacks from bad actors.
It follows Amodei’s visit to Canberra during which he braced the federal government for AI’s rapid progress, warning it is “moving faster than we’d like it”, and posed serious cyber risks.
