Five Eyes Agencies Warn AI Could Dramatically Escalate Cyberattacks Within Months

Artificial intelligence is poised to significantly amplify cyberattacks against governments and major corporations within months, according to a rare joint warning from the intelligence agencies of the Five Eyes alliance. The cyber chiefs of the United States, Britain, Canada, Australia, and New Zealand stated that the rapid advancement of frontier AI models could render long-standing assumptions about digital threats obsolete.

"The timeline is not years, it is months," the agencies warned in a joint statement. They highlighted that frontier AI systems are evolving so quickly that they are anticipated to fundamentally transform both offensive and defensive cyber capabilities.

Recent weeks have seen a surge in warnings regarding the security risks associated with AI. Earlier this month, Gen. Joshua Rudd, head of the National Security Agency, reportedly informed Congress that Anthropic's "Mythos" model had penetrated classified systems rapidly. While the specific claim was later clarified, it underscored growing concerns.

The U.S. administration has also taken actions, including blocking foreign nationals from using Anthropic's "Claude Fable 5" model due to concerns about its power. In response to U.S. export controls, Anthropic temporarily removed its Mythos and Fable models offline to ensure compliance with federal directives. This follows a U.S. ban on G7 countries accessing Anthropic's most advanced AI models, citing national security grounds.

The Five Eyes warning reflects increasing apprehension among Western intelligence officials that the latest generation of AI could lower barriers for malicious actors, increasing the speed and sophistication of cyberattacks. While acknowledging AI's potential to bolster cyber defenses, the agencies cautioned that it can also assist adversaries in identifying vulnerabilities, automating attacks, and exploiting weaknesses faster than organizations can respond.

"AI is not a future consideration — it is already here," the statement read, emphasizing that the technology is shrinking the window between vulnerability discovery and exploitation. This puts increased pressure on businesses and government agencies already struggling with security updates.

The agencies stressed that cyber risk should no longer be treated as solely a technical issue but as a core business risk requiring leadership responsibility. They advised executives to understand cyber risks, empower security leaders, regularly test defenses, and integrate cyber resilience into their business strategy.

Key areas for immediate attention include reducing unnecessary internet exposure, accelerating software patching, replacing outdated legacy systems, tightening access controls, and preparing for inevitable breaches. The agencies also warned that AI systems themselves could introduce new, unknown vulnerabilities, including zero-day exploits.

Rather than relying on single defensive technologies, organizations are urged to adopt multiple layers of protection. The intelligence alliance stressed that cyber incidents should be viewed as inevitable, making preparedness for containment and recovery as crucial as prevention. "Breaches will occur," the statement warned, adding that preparedness helps contain them quickly and prevent escalation into major crises.

Simultaneously, the Five Eyes agencies encouraged organizations to adopt AI-powered defensive tools to counter adversaries who are already leveraging AI. AI can help security operations identify vulnerabilities faster, improve software quality, detect unusual activity, and respond more rapidly to incidents.

"Adversaries are already using AI to move faster and more effectively," the agencies stated. "Defenders must do the same." The warning concluded with a stark message: cyber resilience is a prerequisite for operational survival in the face of rapidly evolving AI threats. As Rep. Andrew Garbarino (R-NY) noted, this advisory underscores concerns that adversarial nations like China are nearing advanced AI capabilities, reinforcing the urgency for federal agencies and critical infrastructure operators to responsibly leverage U.S. models to identify vulnerabilities before they can be exploited.