Five Eyes Alliance Warns AI-Driven Cyberattacks Could Arrive Within Months

The Five Eyes intelligence alliance issued a joint warning on June 22, 2026, saying frontier AI models are advancing fast enough to enable devastating cyberattacks on governments and businesses within months, not years.

The alliance, which includes intelligence and cybersecurity agencies from the United States, United Kingdom, Australia, Canada, and New Zealand, released a statement titled "The AI shift in cyber risk: why leaders must act now."

The agencies said AI lowers the barrier for malicious actors, increasing the speed, scale, and complexity of cyber threats. The window between the discovery of a software vulnerability and its exploitation is shrinking as AI tools become more capable.

The warning followed a U.S. government directive that forced Anthropic to restrict access to its advanced AI models, Mythos and Fable 5, citing national security concerns. Those models demonstrated an advanced ability to identify and exploit software vulnerabilities. A widely circulated claim that Mythos had breached NSA systems was later clarified as an authorized red team test.

The Five Eyes agencies recommended that organizations adopt "secure-by-design" practices, accelerate the patching of known vulnerabilities, reduce attack surfaces, and integrate AI tools into their own security operations to detect threats faster.

The statement explicitly said cyber risk is no longer solely a technical issue. It called on organizational leadership to treat it as a core business risk.

The warning came alongside a May 2026 guidance document on the risks of agentic AI systems, which are models capable of taking independent actions without human oversight.