Five Eyes warns frontier AI models like Mythos pose critical cyber threats
Cyber attack timelines shrink to months as AI-powered hacking evolves.
In a rare joint statement, the Five Eyes intelligence alliance (US, UK, Canada, Australia, and New Zealand) warned that cutting-edge AI models are poised to fundamentally transform offensive cybersecurity. The three-page document, co-signed by agencies including CISA, states that frontier AI models "are anticipated to exceed current industry expectations" and that the timeline for these threats is not years but months. The alert specifically cites advanced models such as Anthropic's Mythos and OpenAI's GPT-5.5-Cyber, which enable rapid, complex, and potentially devastating hacking operations.
The warning comes amid escalating government actions. Earlier this month, Anthropic was forced to disable a version of Mythos after the US government ordered suspension of access for foreign nationals over national security concerns. Concurrently, CISA slashed the deadline for federal agencies to patch critical vulnerabilities from weeks to just three days, explicitly citing the acceleration of AI-driven cyber threats. While the Five Eyes statement reiterates standard cybersecurity advice—patch quickly, limit exposure, use AI for defense—its urgency signals that the window for proactive defense is narrowing rapidly. Defenders are urged to deploy AI to identify weaknesses faster and respond to incidents in real time.
- Five Eyes warns frontier AI models (Anthropic's Mythos, OpenAI's GPT-5.5-Cyber) will transform cyber offense in months, not years.
- Anthropic disabled a Mythos version after US national security order; CISA cut vulnerability patching deadlines to three days.
- Alliance urges defenders to leverage AI for faster weakness detection and incident response, alongside basic security hygiene.
Why It Matters
Governments signal that AI-powered hacking is imminent, forcing organizations to rethink cyber defense timelines and strategies.