Five Eyes Warns AI-Cyber Threats Could Become Reality Within Months

###Bitcoin BTC Five Eyes, the intelligence-sharing alliance of the US, UK, Canada, Australia and New Zealand, warned that cybersecurity threats driven by artificial intelligence could become reality within months and called for immediate action.

"Cyber risk is a core business risk and a leadership responsibility"

Reuters says warning signals rising concern over Anthropic, OpenAI models

The Five Eyes intelligence alliance of the US, UK, Canada, Australia and New Zealand warned that cybersecurity threats driven by artificial intelligence could become reality within months and urged immediate action.

In a joint statement released on July 22, the group said cutting-edge frontier AI models are likely to fundamentally reshape cyber capabilities in both attack and defense beyond current industry expectations. It said the paradigm shift would unfold within months, not years.

The alliance urged national leaders to act now. It called for assessments of AI-related risks, preparedness and accountability; prioritizing basic cybersecurity practices and control systems; fully empowering chief cybersecurity officials with authority and resources; and staying actively engaged with evolving threat indicators and guidance.

The agencies said the response must extend across organizations and society. Cyber risk is no longer just a technical issue but a core business risk and a leadership responsibility, they wrote.

The statement did not cite specific examples or evidence of AI-based cyberattacks.

In the market, the warning is being viewed as an extension of broad security concerns that have persisted since the so-called Mithos shock, which fueled fears that AI could fundamentally disrupt the existing security ecosystem.

Still, the statement carries added weight because intelligence agencies directly warned about cybersecurity risks posed by AI.

Reuters said the warning signals that concern among global intelligence authorities over advanced AI models such as Anthropic's Mithos and OpenAI's GPT 5.5-Cyber has reached a dangerous level.

The statement did not mention any models by name. Still, it is widely viewed as targeting Anthropic's Mithos, which has recently been hit by export-control measures.

On July 12, Anthropic said the US government had issued export-control guidance fully suspending foreign nationals' access to Claude Mithos 5 and Claude Fable 5.

South Korea's government and some companies had recently joined Project Glasswing, which provides access to Mithos. But the latest action by the US administration has effectively constrained their ability to use the model.

Park Sang-kyung, Hankyung.com reporter highseoul@hankyung.com