美国政府批准Anthropic向受信任机构重新开放Mythos 5模型US Government Clears Anthropic to Redeploy Mythos 5 AI Model
美国政府已批准人工智能初创公司Anthropic向大约100家受信任的美国机构重新开放其强大的Claude Mythos 5模型，这些机构包括政府部门和私营企业。此举部分撤销了两周前特朗普政府因国家安全担忧而下达的出口管制指令。当时的指令曾要求Anthropic暂停该模型的访问，并禁止任何外籍人员（包括其外籍员工）进行访问。美国商务部长霍华德·卢特尼克在致Anthropic联合创始人汤姆·布朗的信中表示，政府确信该公司已部署了适当的安全防护措施，以允许受信任的伙伴访问该系统。然而，此次批准并不包括Anthropic的另一个先进模型Fable 5，该模型目前仍处于禁用状态。在此次与特朗普政府的谈判中，联合创始人汤姆·布朗已接替首席执行官达里奥·阿莫代伊主导谈判。此举正值特朗普政府加强对前沿人工智能模型监管之际，旨在防止这些技术被外部军事智能机构滥用。
The U.S. government has allowed artificial intelligence startup Anthropic to restore access to its powerful Claude Mythos 5 AI model for a select group of around 100 trusted organizations, including government agencies and private companies. The decision partially rolls back an export control directive issued two weeks ago over national security concerns, which had instructed the company to suspend all access by any foreign nationals, including its own foreign national employees. In a letter to Anthropic co-founder Tom Brown, U.S. Commerce Secretary Howard Lutnick stated that the government is confident appropriate safeguards are in place for trusted partners to access the system. However, this approval does not extend to Anthropic's other advanced model, Fable 5, which remains restricted. Amid the negotiations with the Trump administration, Brown has taken the lead role for Anthropic, replacing CEO Dario Amodei. The development comes as the administration maintains aggressive oversight of frontier AI models to prevent potential misuse by foreign military intelligence.
01部分解禁：Mythos 5 获准重新部署Partial Relief: Mythos 5 Approved for Redeployment
美国商务部长霍华德·卢特尼克在致Anthropic联合创始人汤姆·布朗的信中表示，他已确定适当的安全防护措施已经到位，允许特定的受信任伙伴访问Claude Mythos 5模型。根据知情人士透露，重新开放将覆盖大约100家组织，其中包括政府机构、私营企业以及许多财富500强公司。这些知情人士表示，恢复访问将有助于确保这些组织能够将该模型用于防御性网络安全目的，而媒体Semafor率先报道了这封信函。
U.S. Commerce Secretary Howard Lutnick stated in a letter to Anthropic co-founder Tom Brown that he has determined appropriate safeguards are in place to permit certain trusted partners to access the Claude Mythos 5 model. According to people familiar with the matter, access to Mythos 5 will be restored to around 100 organizations, including government agencies, private companies, and many Fortune 500 firms. These individuals noted that the restoration of access will help ensure these organizations can utilize the model for defensive cyber purposes, with the outlet Semafor first reporting on the letter.
Anthropic在周五晚间于社交媒体平台X上发表声明，确认政府已通知该公司，其最强大的网络安全模型Mythos 5可以重新部署给一组运营和捍卫关键基础设施的美国组织。美国商务部在周五晚间未立即回应置评请求。这一决定由商务部在信中进行了详细说明，也是双方在经历两周对峙后谈判的一部分。
Anthropic confirmed the decision in a social media post on X on Friday night, writing that the government notified the company that Mythos 5, its strongest cybersecurity model, can be redeployed to a set of U.S. organizations that operate and defend critical infrastructure. The Commerce Department did not immediately respond to a request for comment on Friday night. This decision, detailed in the letter, comes as part of negotiations following a two-week standoff.
02两周对峙背景与出口管制指令Background of the Two-Week Standoff and Export Controls
这场对峙始于本月早些时候，当时Anthropic宣布停用其新推出的Fable 5和Mythos 5模型。该公司采取这一行动是为了遵守美国政府的一项出口管制指令，该指令引用了“国家安全授权”。根据该指令，Anthropic被要求暂停任何外籍人员对这些系统的访问，无论其身处美国境内还是境外。这一限制同样适用于Anthropic自身的外籍员工。
The standoff began earlier this month when Anthropic announced the sudden disabling of access to its newly launched Fable 5 and Mythos 5 models. The company took this action to comply with an export control directive from the U.S. government, which cited national security authorities. Under the directive, Anthropic was instructed to suspend all access to these systems by any foreign national, whether located inside or outside the United States. This restriction also applied to Anthropic's own foreign national employees.
这一政府干预发生在Anthropic公开推出这两款模型的几天后。该公司曾宣传这两款模型在多项行业基准中达到了最先进的水平。其中，Fable 5是该公司首次向公众发布如此先进的模型，并配备了旨在阻止特定高风险领域响应的新安全防护措施。这一突如其来的指令迫使该公司与特朗普政府展开了为期两周的对峙，双方就系统的安全性进行了谈判。
This government intervention occurred just days after Anthropic had publicly introduced the two models, promoting them as state-of-the-art systems across several industry benchmarks. Fable 5 represented the company's first public release of such an advanced model, featuring new safeguards designed to block responses in high-risk areas. The sudden directive forced the company into a two-week standoff with the Trump administration as both sides negotiated the security of the systems.
Anthropic 核心模型发布与监管演变脉络Evolution of Anthropic's Core Models and Regulatory Actions
梳理 Anthropic 核心模型从高估值发布到遭遇紧急出口管制，再到部分解禁的完整监管演变过程。
Traces the regulatory journey of Anthropic's core models from high-valuation release to emergency export controls and partial relief.
Anthropic 在新一轮融资中筹集了 650 亿美元，估值达到 9650 亿美元，成为全球最有价值的 AI 初创公司。
Anthropic raised $65 billion in a funding round, pushing its valuation to $965 billion and becoming the world's most valuable AI startup.
Anthropic 向企业客户和付费订阅用户开放了其最强大的 Mythos 级别 AI 模型 Claude Fable 5。
Anthropic released Claude Fable 5, its most powerful Mythos-class AI model, to enterprise customers and paid subscribers.
为遵守美国政府旨在限制外国国民访问的紧急出口管制指令，Anthropic 停用了其最新发布的 Fable 5 和 Mythos 5 模型。
Following an emergency U.S. export control directive to block foreign nationals, Anthropic disabled access to its Fable 5 and Mythos 5 models.
03谈判进展与未获批准的 Fable 5Negotiation Progress and the Excluded Fable 5 Model
在与特朗普政府的谈判中，Anthropic的谈判主导权发生了变动。根据知情人士透露，双方就系统安全进行了数周谈判，以推动解除对该公司两款顶尖模型的限制。在此期间，联合创始人汤姆·布朗接替了首席执行官达里奥·阿莫代伊，主导与政府的谈判。商务部长卢特尼克的官方信函也是直接寄给布朗的，证实了他的主导角色。
A leadership shift occurred within Anthropic during negotiations with the Trump administration. According to people familiar with the matter, both sides have engaged in weeks of talks regarding system security to resolve restrictions on the company's top two models. During this process, Anthropic co-founder Tom Brown took the lead in negotiations, replacing Chief Executive Officer Dario Amodei. The official letter from Commerce Secretary Howard Lutnick was addressed directly to Brown, confirming his lead role.
尽管Mythos 5取得进展，但当前批准的范围仍然有限。卢特尼克的信函并未允许Anthropic恢复对Fable 5模型的访问，该模型目前依然处于禁用状态。作为该公司首次向公众发布的先进模型，Fable 5配备了旨在阻止高风险领域响应的安全防护措施。虽然该部分协议允许重新部署侧重网络安全的Mythos 5，但针对Fable 5的谈判仍在继续。
Despite progress on Mythos 5, the current approval remains limited. Lutnick's letter did not grant Anthropic permission to restore access to Fable 5, which remains blocked. Fable 5, which was the company's first public release of such an advanced model featuring safeguards for high-risk areas, remains restricted. While the partial agreement allows the redeployment of the cybersecurity-focused Mythos 5, negotiations regarding Fable 5 continue.
04行业背景：前沿AI模型的安全监管趋势Industry Context: Regulatory Scrutiny on Frontier AI Models
针对Anthropic的监管行动反映了特朗普政府对前沿人工智能模型采取的更为严格的监管趋势。由于担忧强大的人工智能系统可能被中国、俄罗斯或其他相关国家的军事智能机构滥用，美国政府对Anthropic及其竞争对手OpenAI的前沿模型发布采取了严厉的监管态度，以保护国家安全利益。
The regulatory actions against Anthropic reflect a broader trend of aggressive oversight by the Trump administration regarding frontier artificial intelligence models. Driven by concerns that powerful AI systems could be misused by military intelligence users in countries of concern, such as China and Russia, the administration has maintained strict scrutiny over model releases from both Anthropic and its rival OpenAI to protect national security interests.
在这种监管压力下，OpenAI在周五早些时候表示，应美国政府的明确要求，该公司将推迟其GPT-5.6模型的全面公开发布。OpenAI没有进行广泛的公开推介，而是将该模型的访问权限限制在少数经过审查的合作伙伴范围内，并已向政府当局分享了这些合作伙伴的具体信息，以确保符合国家安全监管的要求。
Under this regulatory pressure, OpenAI announced earlier on Friday that it is delaying the full public launch of its GPT-5.6 model at the explicit request of the U.S. government. Instead of proceeding with a broad public release, OpenAI has limited access to a small group of vetted partners, and the company has shared the specific details of these approved partners with federal authorities to ensure compliance with national security oversight.
