Five Eyes Warns Frontier AI Is Outpacing Cybersecurity Defences

The Five Eyes intelligence alliance — comprising the United States, United Kingdom, Australia, Canada, and New Zealand — has issued a joint advisory warning that frontier AI models are improving so rapidly they can outpace prevailing cybersecurity defences within months, not years. The advisory, covered by The Straits Times and citing Singapore's partnership in intelligence-sharing frameworks, specifically references the unprecedented vulnerability-finding capabilities of Anthropic's Mythos models as a wake-up call for the global cybersecurity community.

The advisory follows a concrete incident: in April 2026, Anthropic revealed that its cutting-edge Mythos models demonstrated unprecedented abilities to find software vulnerabilities. By June, the company suspended access to Mythos 5 and a restricted variant, Fable 5, after receiving a US government directive banning foreign nationals from accessing the two models. The Five Eyes statement warns that AI lowers barriers for malicious actors while simultaneously increasing the speed and complexity of attacks, making traditional defence timelines obsolete.

The alliance's actionable recommendations include integrating AI tools into existing security operations, hardening legacy systems most vulnerable to exploitation, limiting access to critical systems to reduce blast radius, and adopting a "breaches will occur" mindset focused on rapid containment rather than prevention alone. The advisory makes clear that static perimeter defence is no longer viable in an era where AI-generated attacks evolve faster than human analysts can keep pace.

Why it matters for Singapore: While Singapore is not a formal member of Five Eyes, it maintains close intelligence-sharing partnerships with the alliance and operates one of Asia's most sophisticated cybersecurity ecosystems. The Cyber Security Agency of Singapore (CSA) has been increasingly vocal about AI-driven threats, and this advisory reinforces the urgency of building AI-native defence capabilities. For Singapore's financial hub, where AI is being embedded into everything from fraud detection to trading algorithms, the Five Eyes warning is a timely reminder that cyber defence strategy must evolve at AI speed — or risk becoming obsolete.