AI ‘months away’ from taking down governments – intelligence group

Five Eyes cyber agencies have warned that frontier models could soon transform offensive hacking capabilities

Advanced artificial intelligence models could soon give hackers the ability to cripple governments, businesses, and critical systems, cyber agencies from the Five Eyes intelligence group have warned.

In a rare joint statement published on Monday, cyber security leaders from Australia, the US, the UK, Canada, and New Zealand said frontier AI models are developing faster than expected and are “anticipated to exceed current industry expectations, fundamentally transforming both offensive and defensive cyber capabilities.”

“The timeline is not years, it is months,” the agencies said, adding that “cyber risk can no longer be treated as a purely technical issue. This is a core business risk and leadership responsibility.”

The statement said AI will help improve cyber defense over time, but is also lowering the barrier for malicious actors, increasing the speed and complexity of attacks, while shrinking the window between vulnerability discovery and exploitation.

The agencies urged organizations to strengthen their digital defenses, update outdated software more quickly, limit access to sensitive systems, and prepare for cyberattacks before they happen.

Read more

While the Five Eyes statement did not name any single model or company, the recent debate over AI security has centered on US developer Anthropic, which has faced scrutiny over its latest and most advanced systems.

Earlier this year, the company said one of its flagship models, Mythos, was too powerful to be released to the general public and limited access to a small group of trusted organizations. The company later introduced Fable 5, a more restricted version of the technology, but both models were subsequently taken offline after the US government ordered that foreign citizens be barred from using them, citing national security concerns.

The developments come amid broader warnings from researchers, technology leaders, and security officials that AI capabilities are advancing faster than governments and institutions can adapt.

Experts have increasingly cautioned that systems designed to boost productivity and strengthen cyber defenses could also be used to automate attacks, lower barriers for malicious actors, and amplify the impact of small groups.