Five Eyes intelligence agencies warn that advanced AI models could enable major cyberattacks within months, prompting urgent calls for business and political leaders to act now.

What happened

Australia, the US, the UK, New Zealand, and Canada's signals intelligence agencies issued a rare joint statement warning that frontier AI models are expected to fundamentally transform offensive and defensive cyber capabilities on a timeline of months, not years. The US also blocked foreign nationals from accessing Anthropic's AI models Fable 5 and Mythos 5, while US intelligence agencies retain early access and NSA staff work alongside Anthropic employees.

What to watch

The agencies emphasize that the timeline for frontier AI models to reshape cyber capabilities is measured in months rather than years, signaling an urgent window for organizations to prepare defenses before these capabilities mature.