Five Eyes warns AI will accelerate cyber attacks within months

The intelligence‑sharing alliance Five Eyes – the United States, United Kingdom, Canada, Australia and New Zealand – issued a joint advisory on 22 June 2026 stating that frontier artificial‑intelligence models are set to transform cyber threats. The agencies say AI can discover and exploit software vulnerabilities in weeks or even days, shortening the window between discovery and active exploitation from years to months. The warning stresses that cyber risk is no longer a purely technical issue but a core business risk that senior leadership must manage.

Key recommendations include accelerating patch management, retiring or replacing legacy systems, tightening identity‑and‑access controls, limiting external connections, and deliberately deploying AI‑based security tools to detect vulnerabilities earlier and respond faster. The advisory also notes that the same AI models, such as Anthropic’s Mythos 5 and emerging systems like GPT‑5, can be used by defenders as well as attackers. Boards and executives are urged to treat cyber resilience as an enterprise‑wide responsibility rather than an IT‑only concern.