Cutting-edge artificial intelligence models are advancing so quickly that they could outpace current cybersecurity measures "within months", warned the Five Eyes intelligence alliance.
Security agencies from the UK, US, Australia, Canada, and New Zealand, known as Five Eyes, are urging governments and businesses to act swiftly to adapt to the challenges posed by rapid advancements in artificial intelligence.
The growing threat of AI-assisted hacking has gained attention following claims by the US start-up Anthropic in April that its Mythos models possess unprecedented abilities to identify software vulnerabilities.
"The speed of progress in advanced AI means assumptions about cybersecurity risks could become outdated within months, not years," the alliance stated in a joint communiqué released on Monday.
AI lowers the barriers for malicious actors while increasing the speed and complexity of attacks, the Five Eyes alliance explained in the communiqué.
"Breaches are inevitable. Preparing in advance enables organisations to contain them quickly and prevent them from spiralling into major operational and financial crises," the statement added.
Among its recommendations for bolstering cyberdefences, Five Eyes suggested integrating AI tools into security operations, updating legacy systems, and restricting access to critical systems.
In June, Anthropic suspended access to its two most powerful models, Fable 5 and Mythos 5, following a directive from the US government citing "national security", just three days after their commercial launch.
This intervention by the White House was particularly surprising given its prior efforts to ease AI regulations and discourage other nations from implementing their own rules.
