美国政府发布紧急指令，Anthropic被迫关停最新AI模型Fable 5与Mythos 5Anthropic Disables Latest Fable 5 and Mythos 5 AI Models Following U.S. Directive

美国政府于东部时间周五下午5点21分向人工智能巨头Anthropic下达了一项紧急出口管制指令。该指令明确引用了国家安全机构的授权，要求该公司立即禁止任何外国国民访问其最新且最强大的Fable 5和Mythos 5人工智能模型。该禁令的覆盖范围极为广泛，不仅限制了美国境外的外国国民，也同样适用于美国境内的所有外籍人士，甚至连Anthropic自身雇用的外籍员工也被排除在访问权限之外。

The U.S. government delivered an emergency export control directive to artificial intelligence giant Anthropic on Friday at 5:21 p.m. Eastern Time. Citing national security authorities, the federal mandate ordered the company to immediately block all foreign nationals from accessing its latest and most powerful AI models, Fable 5 and Mythos 5. The scope of the restriction is exceptionally broad, applying to foreign nationals both inside and outside the United States, and explicitly extending to Anthropic's own foreign-born employees.

面对这一突如其来的政府指令，Anthropic表示，为了确保完全合规，公司被迫采取了极端措施，即直接对所有客户停用这两款模型，而不仅仅是针对外国国民。该公司在其周五晚间发布的官方声明中解释称，这一决定的实际效果是必须突然关停Fable 5和Mythos 5的访问服务。不过，Anthropic强调，其旗下的其他人工智能模型并未受到此次管制指令的影响，目前仍保持正常运行。

In response to the sudden government order, Anthropic stated that it had to take the drastic step of disabling access to the models for all of its customers, rather than just foreign nationals, to ensure full compliance. The company explained in its Friday night statement that the net effect of the directive required the abrupt shutdown of Fable 5 and Mythos 5. However, Anthropic clarified that access to all of its other AI models remains unaffected and they continue to operate normally.

Fable 5和Mythos 5在周二刚刚发布，被Anthropic定位为其迄今为止最强大的模型。这两款系统拥有相似的底层技术基础。其中，Fable 5是Anthropic首次向公众发布的如此先进的模型。为了防范潜在风险，Anthropic在Fable 5中引入了额外的安全与防护限制。该公司此前曾警告称，如果没有这些安全防护措施，该模型可能会被滥用并造成严重损害。这些限制措施专门用于阻止该模型在敏感的网络安全、生物和化学领域回答相关问题或提供响应。

Released just days earlier on Tuesday, Fable 5 and Mythos 5 were marketed by Anthropic as its most powerful models to date. While both systems share similar underlying technical foundations, Fable 5 marked the first time the company released such an advanced model to the general public. To mitigate potential risks, Anthropic deployed Fable 5 with additional safety and security restrictions. The company had previously warned that the model could be misused to cause serious damage without these safeguards, which specifically include restrictions designed to block responses on sensitive cybersecurity, biology, and chemistry topics.

相比之下，Mythos 5的发布渠道则更为受限。该模型并未向公众开放，而是仅定向发布给信赖的合作伙伴。Mythos 5和Fable 5均建立在今年4月发布的Claude Mythos Preview模型基础之上。当时，该预览版模型凭借其先进的网络安全能力吸引了华尔街和政府官员的广泛关注。Anthropic曾明确表示，不计划将该模型向公众普遍开放，并作为其名为“Glasswing项目”的网络安全倡议的一部分，将推广范围限制在少数选定的公司内。

In contrast, the rollout of Mythos 5 was far more restricted, with availability limited exclusively to trusted partners. Both models built upon the release of the Claude Mythos Preview in April, which had drawn significant attention from Wall Street and government officials due to its advanced cybersecurity capabilities. Anthropic had previously stated that it did not plan to make that preview model generally available, limiting its deployment to a select group of companies as part of a cybersecurity initiative known as Project Glasswing.

在周五晚间发布的官方声明中，Anthropic向因服务突然中断而受到影响的客户表达了歉意。该公司明确指出，美国国家安全机构在下达该项紧急出口管制指令时，并未提供任何关于其国家安全担忧的具体细节。这使得该公司在无法获得具体安全威胁解释的情况下，不得不采取全面关停的极端合规措施。

In its official statement released late Friday evening, Anthropic apologized to its customers for the sudden disruption of service. The company specifically pointed out that U.S. national security authorities failed to provide any specific details regarding the national security concerns that prompted the emergency export control directive. This lack of specificity left the artificial intelligence developer to implement a complete shutdown of its premier models without a detailed explanation of the alleged risks.

与此同时，Anthropic对政府此次干预的程序合理性提出了公开质疑。该公司重申其一贯的公开立场，即政府确实应当拥有阻止不安全部署的能力，但这种干预必须作为法定程序的一部分，且该程序应当遵循透明、公平、清晰以及基于技术事实的原则。Anthropic直言不讳地指出，政府周五傍晚采取的这一紧急行动并不符合上述原则。

Furthermore, Anthropic openly criticized the procedural nature of the government's sudden intervention. The company outlined its stance that while the federal government should possess the authority to block unsafe AI deployments, such actions must be executed as part of a statutory process that remains transparent, fair, clear, and grounded in technical facts. Anthropic stated that the emergency directive issued on Friday evening did not adhere to these fundamental principles.

此次紧急关停在整个科技行业引发了广泛关注。这是首次有领先的人工智能公司因为联邦政府的直接干预，而被迫将其已经公开部署的人工智能模型下线。这一行动不仅打破了政府与人工智能开发商之间的常规互动模式，也为未来监管机构如何介入先进人工智能系统的分发与运行树立了先例。

The emergency shutdown has drawn significant attention across the technology sector, marking the first time a leading artificial intelligence company has been forced to take a publicly deployed model offline due to direct federal government intervention. This unprecedented action disrupts the typical dynamics between AI developers and regulators, establishing a major precedent for how the government may intervene in the distribution of advanced systems.

在此次事件发生的同时，Anthropic与特朗普政府之间还存在着另一起独立的法律纠纷。双方目前正陷入一场诉讼中，起因是政府此前下达了一项禁止联邦机构使用该公司人工智能工具的命令。这一既有的诉讼背景，使得此次因国家安全指令而紧急关停最新模型的事件，让该公司与联邦监管机构之间的关系变得更加复杂。

This confrontation unfolds alongside an existing, separate legal dispute between Anthropic and the Trump administration. The two parties are currently involved in an ongoing lawsuit over a prior government order that banned federal agencies from using the company's AI tools. This ongoing litigation adds a layer of regulatory tension to the sudden national security directive that halted the company's newest models.