Overview
- The leaders of the Five Eyes cyber agencies issued a joint statement on Monday saying frontier AI is accelerating cyber threats and shortening the window between vulnerability discovery and exploitation from years to months.
- The agencies set out immediate steps for executives to act on, including cutting internet-exposed systems, accelerating patching, retiring unsupported legacy systems, tightening identity and access controls, and regularly testing incident-response plans.
- They warned AI is dual use: while it lowers barriers for attackers by enabling faster, more complex exploits, it can also help defenders detect weaknesses earlier, monitor unusual behaviour, and speed incident response when used deliberately.
- The advisory follows recent industry moves, including Anthropic’s June pause on access to its Fable 5 and Mythos 5 models after testing by Britain’s AI Security Institute found high exploitability, a development cited by the agencies as a warning sign.
- The statement builds on a U.S. executive order from early June and is likely to push boards, regulators and insurers to demand faster fixes, raise scrutiny of cyber resilience, and increase demand for defensive AI tools and basic hygiene measures.