US Government Imposes Export Controls on Two Anthropic AI Models, Restricting Global Access

On June 12, 2026, the US government invoked national security authority to impose export controls on Anthropic's newly released AI models, Fable 5 and Mythos 5, banning access by any foreign nationals, including Anthropic's foreign employees. To comply, Anthropic globally disabled both models just 72 hours after their June 9 release. This marks the first time the US government has used export controls on deployed commercial AI models.

Background

On June 9, Anthropic released Fable 5 (for developers) and Mythos 5 (for government and allied enterprises), claiming they outperformed existing models on most benchmarks.
Prior to release, Anthropic conducted thousands of hours of red-teaming and collaborated with the US government and UK AISI, but the government had unsuccessfully attempted to block the release.
At 5:21 PM ET on June 12, US Commerce Secretary Howard Lutnick sent a letter to Anthropic CEO Dario Amodei announcing the export control order.

Key Details

Scope: Prohibits export to any region outside the US and bans access by all foreign nationals within the US, including Anthropic's foreign employees.
Anthropic's Response: Unable to precisely distinguish user nationality, it disabled Fable 5 and Mythos 5 for all users globally. Other Claude models remain unaffected.
Trigger: According to Axios, another company claimed to have successfully jailbroken the Mythos model, raising government concerns. Anthropic stated that the government provided only verbal evidence of a narrow, non-general jailbreak that also exists in other models (e.g., GPT-5.5).
Anthropic's Position: Disagrees with recalling models based on a narrow jailbreak, arguing that such a standard would halt deployment of all frontier models. The company is working to restore access.

Reactions and Impact

Anthropic: Issued a statement saying it complies with the legal order but believes it is based on a misunderstanding and criticized the lack of transparent, fair legal process.
Amazon AWS: At Anthropic's request, revoked all user access to Fable 5 and Mythos 5.
Industry Impact: This is the first time in AI history that a deployed commercial model has been forcibly recalled by the government. Anthropic, previously blacklisted by the Pentagon as too dangerous for government use, now faces Commerce Department controls for being too dangerous for foreign use.
User Impact: Existing Fable 5 sessions error out; new sessions default to other models or Opus 4.8. Developers must update integrations to switch models.

Data and Timeline

Model release: June 9, 2026
Government order received: June 12, 2026, 5:21 PM ET
Global disablement: Immediately upon receipt
Model lifespan: Approximately 72 hours