Five Eyes agencies warn AI could reshape cyber attacks within months

Signals agencies from Australia, the US, the UK, New Zealand and Canada issued a rare joint warning that powerful AI models capable of substantially escalating cyber attacks against governments and businesses are expected within months. The Five Eyes agencies said AI will improve cyber defence over time, but also accelerates the speed, scale and sophistication of threats, requiring leaders to “act now”.

The statement followed a Trump administration decision to block “foreign nationals” from using Anthropic models including Fable and Mythos, citing national security advice. Anthropic’s Fable 5 is described as a more community-friendly version of Mythos, a model designed to detect vulnerabilities in cyber systems and limited to vetted organisations because of misuse concerns.

Olivia Shen, a national security and AI expert at the University of Sydney’s United States Studies Centre, said the latest models are especially capable of generating exploits, lowering the barrier for malicious actors. She warned that similar systems could be under development by China, other states, actors or companies. Australia’s Albanese government has signed Anthropic to its national AI plan under a non-binding memorandum focused on sharing progress and promoting safety while pursuing light-touch regulation.