Five Eyes agencies warn that AI-powered cyberattacks are months away as the US moves to lock down frontier models

The Five Eyes intelligence alliance has warned that frontier AI models capable of dramatically escalating cyberattacks are months away from broad accessibility, not years. The warning arrives alongside the US government's unprecedented export ban on Anthropic's Fable 5 and Mythos 5 models, a move that cut allied governments off from frontier access while raising unresolved questions about enterprise liability and startup resilience.