美国商务部解除对Anthropic先进AI模型的出口管制US Lifts Export Controls on Anthropic's Most Powerful AI Models

美国人工智能公司Anthropic于周二宣布，美国商务部已正式解除针对其Claude Fable 5和Mythos 5模型的出口管制指令。这一决定标志着该公司与特朗普政府之间持续数周的对峙得以化解。Anthropic通过社交媒体平台X发文，对用户的耐心等待以及所有协助重新部署模型的人员表示感谢，并确认将从周三（7月1日）开始恢复全球用户对这些先进技术的访问权限。

Artificial intelligence company Anthropic announced on Tuesday that the U.S. Department of Commerce has officially lifted the export controls previously imposed on its Claude Fable 5 and Mythos 5 models, successfully resolving a weeks-long standoff with the Trump administration. In a statement shared on the social media platform X, the company expressed gratitude to its users for their patience and thanked everyone who assisted in the effort to redeploy the models. Anthropic confirmed that global access to these advanced systems will begin restoring on Wednesday, July 1.

根据Anthropic发布的声明，Claude Fable 5模型将重新在Claude.ai平台和Claude Code上向全球用户开放。此次重新上线不仅解决了此前因出口管制导致的全球服务中断问题，也让国际用户和开发人员能够重新使用该公司的旗舰AI技术。通过与美国政府的沟通与合作，双方最终就技术安全和合规达成了共识，从而使这一最先进的AI模型得以重回全球市场。

Under the redeployment plan, the Claude Fable 5 model will once again be accessible to global users through the Claude.ai platform and Claude Code. The lifting of the export restrictions marks a significant shift, allowing international clients and developers to resume utilizing Anthropic's most capable artificial intelligence technologies after weeks of forced suspension. The resolution follows close coordination between the company and federal officials to address regulatory and security concerns.

这场对峙始于今年6月中旬。在Anthropic向公众发布Claude Fable 5模型仅几天后，特朗普政府便发布了出口管制指令，迫使该公司于6月12日将该模型及其另一款模型Mythos 5紧急下线。政府高级官员当时指出，这些模型存在严重的网络安全风险，并认为Anthropic的领导层未能充分重视官方的担忧。此外，政府官员还担心这些模型可能会被黑客用于恶意网络攻击或其他危害国家安全的活动。

The standoff began in mid-June, just days after Anthropic released Claude Fable 5 to the public. On June 12, the Trump administration issued an export control directive citing national security authorities, forcing the company to abruptly take both Fable 5 and its sibling model, Mythos 5, offline. Senior administration officials stated that the models posed severe cybersecurity risks, expressing concern that they could be exploited by hackers for cyberattacks or other nefarious purposes, and asserted that Anthropic's leadership had not sufficiently recognized these government concerns.

根据当时政府的宽泛指令，Anthropic被要求禁止任何外籍人员访问这些模型，无论其身处美国境内还是境外，这甚至包括了Anthropic自身雇用的外籍员工。由于该限制范围极广，Anthropic表示这实际上等同于要求其完全禁用这些模型。这一出口管制指令随即引发了外界对人工智能主权和保护主义的广泛担忧，也导致该公司最先进的技术部署被迫陷入停滞。

Under the broad federal order, Anthropic was instructed to suspend all access to the models by any foreign national, whether located inside or outside the United States, a restriction that explicitly included Anthropic's own foreign national employees. Anthropic stated that this sweeping mandate effectively required it to disable the models entirely. The abrupt suspension of access sparked immediate concerns over artificial intelligence sovereignty and protectionism, freezing the deployment of the company's most powerful technologies.

美国商务部长豪尔德·卢特尼克在致Anthropic联合创始人汤姆·布朗的信中，详细阐述了取消出口管制许可的具体条件。卢特尼克表示，在过去几周里，其团队与Anthropic开展了密切合作，对Claude Fable 5模型进行了深入的分析与评估，以确保美国政府内部在安全立场上保持一致，并巩固美国在人工智能领域的领导地位。这一密切的协调与合作，最终促成了商务部对该模型的批准并允许其重新上线。

U.S. Commerce Secretary Howard Lutnick detailed the specific conditions for lifting the export restrictions in a letter sent to Anthropic co-founder Tom Brown. Lutnick stated that over recent weeks, his team worked closely with Anthropic to analyze and approve the Fable 5 model, ensuring alignment across the federal government while strengthening American leadership in artificial intelligence. This close coordination and cooperation to address the risks associated with the covered models ultimately paved the way for the Commerce Department's approval.

根据双方达成的共识，Anthropic同意采取一系列主动的安全防范措施，从而无需再申请出口许可。这些承诺包括主动检测并解决与模型相关的潜在安全风险、与美国政府合作制定未来新模型的安全标准，以及在发现任何恶意活动时及时向政府通报。商务部正是基于这些在技术安全和合规保障上的具体承诺，决定取消此前对这两款先进模型实施的严格限制。

Under the newly reached consensus, Anthropic will no longer require an export license after agreeing to a series of proactive security commitments. The company has pledged to proactively detect and address safety and security risks associated with its models, collaborate with the government on establishing standards for upcoming models, and promptly inform federal authorities of any detected malicious activity. These commitments to ongoing risk mitigation and regulatory compliance satisfied the government's national security concerns, leading to the removal of the export controls.

在重新部署的具体方案中，Claude Fable 5模型将重新向全球用户开放，并被纳入特定订阅计划的使用额度中。Anthropic表示，在7月7日之前，该模型将为Pro、Max、Team以及部分企业计划的用户提供最多占每周使用限制50%的额度。这一安排旨在让这些付费和企业级用户在模型重返市场初期，能够有保障地体验到这款最先进模型的各项功能。

Under the specific redeployment plans, Claude Fable 5 will return to global users with temporary usage allocations. Anthropic stated that through July 7, the model will be included for up to 50% of weekly usage limits for subscribers of its Pro, Max, and Team tiers, as well as selected enterprise plans. This temporary allocation structure ensures that premium and corporate clients can immediately begin reintegrating the advanced model into their workflows during its initial return phase.

与此同时，另一款模型Mythos 5的部署也在逐步推进。在获得美国政府于6月26日的批准后，Anthropic已经恢复了部分美国机构对Mythos 5的访问权限。该公司表示，将继续与美国政府合作，通过其名为“Glasswing”的网络安全计划，向更多的国内和国际合作伙伴扩大Mythos 5的访问范围。该计划是一项专门的防御性安全倡议，旨在为选定的机构提供接触先进人工智能模型的通道，以便开展防御性安全测试。

Meanwhile, the deployment of the sibling model, Mythos 5, is progressing through a more controlled framework. Following government approval granted on June 26, Anthropic has already restored access to Mythos 5 for select U.S. organizations. The company plans to continue working with federal authorities to expand access to Mythos 5 for additional domestic and international partners through its Glasswing program, a cybersecurity initiative designed to provide designated organizations with access to advanced AI models for defensive security testing.