An international intelligence alliance has warned that artificial intelligence systems capable of enabling large-scale cyberattacks that could overwhelm government and corporate defenses may emerge in a matter of months rather than years.
The Five Eyes alliance—comprising the United States, United Kingdom, Canada, Australia, and New Zealand—called on governments and private sector leaders to urgently strengthen their cyber defenses against rapidly evolving AI-driven threats.
The rare warning follows recent actions by the Trump administration directing AI firm Anthropic to restrict foreign access to its most advanced models, underscoring growing Western concern over the security risks posed by frontier AI technologies.
In a joint statement released on Monday, the intelligence agencies said: “Frontier AI models are anticipated to exceed current industry expectations, fundamentally transforming both offensive and defensive cyber capabilities. The timeline is not years, it is months.”
They added that the fast-changing AI landscape is significantly reshaping cyber risk, stressing the need for immediate action to avoid falling behind emerging threats.
Security experts and AI industry leaders have increasingly raised concerns that advanced models could lower the barrier for cybercriminals, enabling faster, more sophisticated, and harder-to-detect attacks. The Five Eyes agencies warned that this could affect not only governments and large corporations but also small and medium-sized businesses globally.
The US government’s reported directive involving Anthropic’s advanced models—internally known as Mythos 5 and Fable 5—has been described as one of the most sweeping interventions yet in response to AI capability risks. The Mythos model in particular reportedly raised cybersecurity concerns due to its ability to identify system vulnerabilities, while reports suggest concerns also emerged over potential “jailbreak” methods affecting the Fable model.
Anthropic has reportedly been in discussions with US authorities to address the issue.
To mitigate risks, the alliance urged organizations to modernize outdated systems, apply security patches regularly, and tightly control access to critical infrastructure. At the same time, they noted that AI can also strengthen cybersecurity defenses.
“Organizations that integrate AI tools into their security operations can detect vulnerabilities earlier, improve software quality, monitor unusual behaviour, and respond faster to incidents,” the statement said.
Separately, dozens of cybersecurity experts, AI founders, and business leaders have called for a more transparent, scientific approach to AI risk assessments, arguing that stronger collaboration is needed to help organizations identify and fix vulnerabilities before they can be exploited by attackers.
