五眼联盟（美、英、加、纽、澳）的网络安全主管联合警告，西方国家的对手可能在几个月内开发出利用人工智能（AI）的网络攻击，进而摧毁政府与企业的防御系统。尽管目前西方在商业AI的发展与军事集成上仍占优势，但这一领先地位可能难以持久。

美国政府近期已下令限制对手国家获取如Anthropic开发的最先进AI模型，视其为安全威胁。同时，网络安全专家已经观测到由AI设计的更复杂攻击，例如Google威胁情报小组曾拦截到疑似由俄罗斯黑客利用AI自动调整并发起的漏洞攻击。

五眼联盟网络安全主管呼吁企业主动采用AI防御工具来强化自身安全，以应对这场防御者与攻击者之间的AI军备竞赛。他们指出，拖延采用AI防御的组织将面临日益增长且本可避免的网络风险。

The cyber chiefs of the Five Eyes intelligence alliance (US, UK, Canada, New Zealand, and Australia) have jointly warned that adversaries could develop AI-powered cyberattacks within months that could overwhelm the defenses of governments and companies. Although the West currently holds an advantage in commercial AI and military integration, this lead may not last long.

The US government recently restricted foreign access to advanced AI models like those developed by Anthropic, classifying their export as a security risk. Meanwhile, cybersecurity experts have already encountered highly sophisticated AI-engineered threats, such as a suspected Russian attack blocked by Google that used AI to automatically adapt and exploit vulnerabilities.

The Five Eyes cyber chiefs urged companies to proactively adopt AI tools to strengthen their cybersecurity defenses, framing the situation as an active AI arms race between targets and adversaries. They warned that organizations delaying the integration of AI into their security operations will face growing and avoidable risks.