```
In a move that signals a major escalation of government oversight into the core of artificial intelligence development, the US Commerce Department has blocked Anthropic from releasing two of its latest advanced AI models.
The directive, reported by Security Affairs, was issued via a formal letter sent to Anthropic on Friday, June 12, at 5:21 PM Eastern Time. The document was signed by Commerce Secretary Howard Lutnick and drafted in coordination with the Bureau of Industry and Security (BIS), the agency that typically oversees technology export controls. It orders the company to halt the development and release of its Mythos 5 and Fable 5 models.
Anthropic has objected to the decision, arguing that the Commerce Department has failed to provide a clear, transparent, or evidence-based technical rationale for singling out these specific systems. This challenge underscores a growing tension in AI policy: the need for effective regulation versus the risk of opaque, unpredictable government intervention.
A Shift from Hardware to Model Controls
The action represents a significant departure from the US government's previous strategy for managing advanced AI capabilities, which has largely centered on restricting the export of high-performance semiconductors—the hardware used to train large models. By intervening directly in the lifecycle of specific AI systems, regulators are establishing a new "model-level" regulatory perimeter.
This shift brings federal authority directly into the development pipeline of US-based AI companies, moving beyond controlling inputs to controlling outputs.
A Precedent in the Making
The case raises urgent questions for the entire AI industry. If the government can block model releases without public disclosure of its technical assessment, it could create uncertainty that chills innovation, particularly among safety-focused researchers who may now be unsure where regulatory boundaries lie.
The precedent set here will be closely watched internationally. As the European Union, United Kingdom, and other jurisdictions craft their own AI governance frameworks, the US approach could either serve as a template for direct state control or become a cautionary tale about regulatory overreach.
The coming weeks will reveal whether Anthropic pursues a legal challenge, negotiates with regulators, or complies. Its response, and the government's subsequent actions, will help define whether the emerging era of AI oversight is built on transparent, technically grounded rules or on ad-hoc, discretionary interventions.
美國商務部已阻止人工智能公司 Anthropic 發布其兩款最新先進的 AI 模型，此舉標誌著政府對人工智能發展核心的監督力度顯著升級。
據《Security Affairs》報道，這項指令於東部時間 6 月 12 日星期五下午 5 時 21 分，透過一封正式信函送達 Anthropic。該文件由商務部長霍華德·盧特尼克簽署，並與通常負責技術出口管制的工業與安全局（BIS）協調起草。信中命令該公司停止開發及發布其 Mythos 5 與 Fable 5 模型。
Anthropic 對此決定提出異議，認為商務部未能提供清晰、透明或基於證據的技術理由，來說明為何單獨針對這些特定系統。此一挑戰突顯了 AI 政策中日益加劇的緊張關係：有效監管的需求與不透明、不可預測的政府干預風險之間的矛盾。
從硬件控制轉向模型層級管控
這項行動標誌著美國政府管理先進 AI 能力的策略出現重大轉變。此前，策略主要集中於限制高性能半導體（即用於訓練大型模型的硬件）的出口。如今，監管機構直接介入特定 AI 系統的生命周期，從而確立了一個新的「模型層級」監管邊界。
這一轉變使得聯邦權力直接深入美國 AI 公司的開發流程，其管控範圍從「控制輸入」延伸至「控制輸出」。
正在形成中的先例
此案例為整個 AI 行業提出了緊迫的問題。如果政府可以在不公開披露其技術評估的情況下阻止模型發布，可能會帶來不確定性，從而抑制創新，特別是對於那些專注安全的研究人員而言，他們現在可能無法確定監管的邊界何在。
此處所樹立的先例將受到國際社會密切關注。隨著歐盟、英國及其他司法管轄區制定各自的 AI 治理框架，美國的做法既可能成為國家直接控制的範本，也可能成為監管過度的警示故事。
未來數週將揭曉 Anthropic 會尋求法律挑戰、與監管機構談判，還是選擇遵從。其回應及政府的後續行動，將有助於定義新興的 AI 監督時代，究竟是建立在透明、有技術依據的規則之上，還是依賴於臨時、酌情進行的干預措施。