US order halts access to Anthropic’s newest Claude models

Anthropic suspended access to its latest Claude models, Fable 5 and Mythos 5, on June 12 after receiving a US government “export control directive” barring use by anyone who is not a US national. Mythos is described as Anthropic’s frontier model, while Fable uses the same basic model with added safeguards intended to prevent cybersecurity use.

The government has not publicly explained the directive, but Anthropic believes officials reacted to evidence of a jailbreak that could bypass Fable’s protections and extract information useful for cyberattacks. Anthropic says the research behind the directive appears to have come from Amazon engineers, while a separate researcher using the pseudonym “Pliny the Liberator” published what they identified as Fable 5’s full system prompt within 48 hours of release.

The dispute follows escalating tension between Anthropic and the Trump administration over AI regulation, semiconductor export policy, and Pentagon access to Anthropic models for domestic surveillance and fully autonomous weapons systems. The broader challenge is that frontier AI systems remain difficult for governments, companies, and the public to assess, making safety rules and guardrails hard to rely on as capabilities advance.