Five Eyes alliance warns frontier AI cyber threats are imminent

The Five Eyes intelligence alliance has issued a stark warning that frontier artificial intelligence models are fundamentally transforming offensive cyber capabilities, with officials stating that the timeline for these threats is measured in months rather than years. In a three-page statement released on Monday, the alliance comprising the United States, United Kingdom, Canada, Australia and New Zealand cautioned that cutting-edge technology is poised to supercharge hacking operations, necessitating urgent action from defenders.

The statement, which was described as light on technical detail, largely restated core cybersecurity advice while urging organisations to utilise AI to strengthen their own defences. Officials advised defenders to identify weaknesses more rapidly and respond to incidents with greater speed, while reiterating standard protocols such as swiftly patching faulty software and keeping systems offline unless strictly necessary.

Specific concerns were highlighted regarding advanced models, including Anthropic’s Mythos and OpenAI’s GPT-5.5-Cyber. These systems are believed to enable users to quickly execute complex and potentially devastating hacks, raising alarms among intelligence officials who fear these tools will exceed current industry expectations in the immediate future.

The warning follows a series of recent regulatory and operational moves by US authorities. Earlier this month, the US government ordered Anthropic to suspend access to certain AI models for foreign nationals over alleged national security concerns, prompting the company to disable a version of the Mythos model. This action underscores the growing tension between technological innovation and national security protocols within the alliance.

Concurrently, the US Cybersecurity and Infrastructure Security Agency (CISA), which was among the co-signatories of the joint statement, reduced the deadlines imposed on government officials to remediate serious digital vulnerabilities to just three days. CISA cited the accelerating threat posed by AI-driven attacks as the primary driver for this significant tightening of operational timelines.

The alliance’s intervention signals a shift in how intelligence agencies view the intersection of generative AI and cyber warfare. By framing the threat as an immediate, month-scale challenge rather than a long-term risk, the Five Eyes nations are pushing for a more aggressive posture in both defensive technology adoption and vulnerability management across their respective networks.