Five Eyes warns AI could outpace cyber defences within months

WASHINGTON: The Five Eyes intelligence alliance has warned that the most advanced artificial intelligence systems are developing fast enough to surpass current cybersecurity practices within months, raising concerns about the speed and scale of AI-driven cyber threats.

In a joint advisory dated Monday, the security agencies of the United States, Britain, Australia, Canada and New Zealand said governments and businesses need to move quickly as AI capabilities advance. The alliance said rapid progress in frontier AI means assumptions used to assess cyber risk may become obsolete far sooner than many organisations expect.

The advisory said AI is making it easier for malicious actors to carry out attacks while also increasing how quickly and how intricately such attacks can unfold. The agencies urged institutions to strengthen their readiness, warning that cyber incidents are inevitable and that preparation is essential to contain damage and avoid wider operational and financial disruption.

“The rapid pace of frontier AI development means cyber risk assumptions can become outdated in months, not years,” said a joint statement dated Monday.

The Five Eyes agencies also said organisations should improve their defences by bringing AI tools into security operations, modernising outdated systems and restricting access to critical infrastructure and systems, among other measures.

The warning comes after US startup Anthropic said in April that its advanced Mythos models had shown unprecedented ability to identify software vulnerabilities. That development has intensified attention on the possibility of AI-enhanced hacking.

Anthropic this month suspended access to Mythos 5 and a restricted model known as Fable 5 in response to a US national security order. The company said that only days after launching Fable 5 publicly, it received a government directive barring all foreign nationals from using the two models.

The move is notable because the White House has otherwise supported easing AI regulation and has also sought to prevent US states from setting their own rules on the technology.