Anthropic Pulled Its Most Advanced AI Models Worldwide After a US Government Order

What Happened

The US Commerce Department ordered Anthropic to block foreign access to its most advanced AI models on June 13, and within hours the company pulled the plug globally. The directive instructed the company to suspend access for any foreign national, whether based overseas or physically present in the United States, a sweep broad enough to include Anthropic's own employees who hold foreign citizenship.

Fable 5 is the first publicly available version of Anthropic's Mythos model, built for software engineering, knowledge work, and vision, but shipped with hard safety limits that block responses in high-risk areas like cybersecurity and biology, falling back to Claude Opus 4.8 instead. Mythos 5 is the same underlying model without those limits, available only to a small group of vetted customers.

For years, US export controls have focused on the chips and tools that power AI rather than on restricting foreign access to AI itself. Anthropic says the government gave it only "verbal evidence of a potential narrow, non-universal jailbreak." A non-universal jailbreak is a workaround that can extract some sensitive information in specific circumstances but cannot broadly unlock a model's capabilities. Anthropic says the technique in question reveals vulnerabilities that other publicly available models can already find without any workaround at all.

The company is complying but is pushing back publicly. It argues that if a narrow, partial jailbreak is sufficient grounds to recall a commercial model, no frontier AI company could ever ship a new product again.

The Bureau of Industry and Security has historically focused on hardware, specifically chips and semiconductors. Applying export control logic to model weights and API access is genuinely new territory, and the directive suggests the US government is starting to treat advanced AI models more like controlled technology than open software.

What This Means for You

If you are running workloads on Fable 5 or Mythos 5 through Anthropic's API or through AWS Bedrock, those calls are failing right now. Anthropic's other models, including Opus 4.8, remain fully operational. Switch any production pipelines to Opus 4.8 or another available Claude model while the suspension holds. Do not wait for an automatic restoration: Anthropic has not given a timeline for when, or whether, the government will lift the order.

Further reading

• TechCrunch: Anthropic's Claude Fable 5 is a version of Mythos the public can access

• Business Standard: Anthropic halts foreign access to advanced AI models on US government's directive

• Al Jazeera: US asks Anthropic to block global access to top AI models

• Crypto Briefing: Mark Carney warns of risks from US export ban on Anthropic AI models