Five Eyes warn frontier AI could outpace cybersecurity defenses within months

The Five Eyes intelligence alliance said frontier AI models could outstrip current cybersecurity defenses within months, urging governments and businesses to accelerate preparations. After highlighting the risks of AI-enhanced hacking, the agencies said organizations should modernize systems, limit access to critical infrastructure and incorporate AI tools into security operations. Anthropic then suspended external access to Mythos 5 and a restricted version, Fable 5, to comply with a U.S. national security order that bars foreign nationals from using the models. The move underscores a shift from broad AI governance principles to direct, operational intervention in a rare instance of urgent enforcement in the sector.