The Five Eyes intelligence alliance, comprising the US, UK, Canada, Australia, and New Zealand, issued a joint statement on June 22, 2026, warning that frontier AI models will outpace current cybersecurity norms within months, drastically increasing the speed, scale, and complexity of cyberattacks [1, 2, 3, 4, 5, 6, 7, 8, 9, 10]. They said, "The rapid pace of frontier AI development means cyber risk assumptions can become outdated in months, not years. Breaches will occur. Preparedness helps you contain them quickly and prevent escalation into major operational and financial crises" [1].

The statement urged governments and businesses to strengthen cyber defenses using AI, patch vulnerabilities quickly, and limit access to critical systems to contain breaches [1, 3, 4, 5, 6, 7, 8, 9, 10]. US Cybersecurity and Infrastructure Security Agency (CISA) responded by shortening deadlines for officials to patch serious digital vulnerabilities to three days from a longer standard period, citing the accelerated threat posed by AI-driven attacks [3, 4, 5, 7].

Anthropic, a US AI startup, has demonstrated advanced AI models—Mythos and Fable 5—with unprecedented ability to identify software vulnerabilities and accelerate complex cyber attacks, raising national security concerns [1, 2, 3, 4, 5, 6, 7, 8, 9, 10]. As a result, the US government ordered Anthropic in June to suspend foreign nationals' access to these models [1, 2, 3, 4, 5, 6, 7, 8, 9, 10].

Cybersecurity experts suspect countries such as China, Russia, and North Korea are developing or using AI-enhanced hacking tools, though the Five Eyes statement did not explicitly name adversaries [2, 5, 6]. China's AI progress is notable, with Zhipu Technology launching GLM-5.2 on June 17, an open-source large language model ranked among the world's top three. Its stock price increased approximately 20 times since January 2026, with a market capitalization of about HKD 1.1 trillion (USD 140 billion) as of June 22 [11, 12].

Former US officials warn of dangerous complacency amid a rapidly closing AI technology gap. David Sacks, a former White House AI official, said, "The US lead in AI technology may only remain six to nine months" [11], while Alex Stamos, former Facebook cybersecurity chief, warned against assuming US AI superiority, stating, "That arrogance is dangerous as China may already be developing very competitive AI technology privately" [11]. AI and national security expert Olivia Shen noted, "The next powerful AI model like Mythos or Fable could soon emerge and global attention can't remain fixed on just the publicly known AI products" [6].

In April 2026, Anthropic announced its Mythos models with new capabilities identifying software flaws [1]. Days before the Five Eyes warning, China's Zhipu launched GLM-5.2, intensifying the AI arms race [11, 12]. US agencies accelerated protective measures in June, including access restrictions and accelerated patching deadlines [1, 3, 4].

The US White House Technology Policy Office anticipates a quantum computing plan by 2028 to help defend government systems against future quantum-enabled cyber threats [2]. Meanwhile, CISA’s new three-day patch deadline remains in effect as governments act to limit damage from evolving AI-driven cyberattacks [3, 4, 5, 7].