美国商务部近日以国家安全为由，发布命令限制非美国公民使用 Anthropic 最新的 AI 模型 Fable 和 Mythos，迫使该公司暂停最新工具，此举给予 Anthropic 仅 90 分钟的合规时间，引发了整个 AI 行业的警惕，业界担忧缺乏明确且一致的监管方式恐阻碍关键技术的发展。

该命令源于亚马逊研究人员对 Fable 模型潜在漏洞的警告，认为其可能被用来规避安全限制并获取其他软体漏洞的资讯；然而，Anthropic 坚称细微的「越狱」漏洞不足以成为召回面向数亿用户商业模型的理由，并指出竞争对手的系统亦具备类似能力，甚至质疑此举背后可能存在政治动机。

华盛顿的强硬手段引发了产业动荡与盟友的不安，数十位网络安全高层联名警告此举削弱了防御者的力量并创造市场不确定性，同时也加剧了欧洲等地区减少对美技术依赖、提升技术自主权的压力；专家呼吁应仿照联邦航空管理局（FAA）成立独立机构，建立既能保障安全又不阻碍创新的智慧监管机制。

The US Department of Commerce recently issued an order restricting the use of Anthropic's latest AI models, Fable and Mythos, by non-US nationals due to national security concerns, forcing the company to suspend its newest tools. The move, which gave Anthropic only 90 minutes to comply, has raised alarms across the AI industry over fears that the lack of a clear and consistent regulatory approach could hinder the development of this crucial technology.

The restriction reportedly followed warnings from Amazon researchers about potential vulnerabilities in Fable that might allow guardrails to be bypassed and security flaws in other software to be exposed. Anthropic argued that a narrow "jailbreak" did not justify recalling a commercial model deployed to hundreds of millions of users, noting that competitor systems possessed similar capabilities and suggesting that the government's sudden decision might be politically motivated.

Washington's heavy-handed action has unsettled the industry and US allies alike, prompting cybersecurity executives to warn that it deprives defenders of top models while creating market uncertainty, and intensifying pressure in Europe to reduce reliance on US technology. Experts advocate for intelligent regulation, suggesting an independent body similar to the Federal Aviation Administration (FAA) to strike a balance between safeguarding national security and fostering technological innovation.