Anthropic Suspends Access to Fable Model Following U.S. Government Directive

Anthropic, an artificial intelligence company, has suspended access to its Fable 5 and Mythos 5 models worldwide following a directive from the U.S. government. The order, issued on June 12, 2026, required the company to restrict access to these models for all foreign nationals, including foreign-national employees, due to national security concerns ^[1][2].

The U.S. Commerce Department confirmed the issuance of the export control directive, which was delivered to Anthropic at approximately 5:21 p.m. ET ^[3]. In response, Anthropic disabled access to both models for all users globally, stating that selective enforcement was not feasible ^[1]. Access to other Anthropic models remains unaffected ^[1].

The directive was reportedly prompted by a report from Amazon to the White House about a jailbreak vulnerability in the Mythos model, which led to urgent government action ^[4]. Anthropic stated that the government provided only verbal evidence of a potential narrow, non-universal jailbreak involving the model reading a codebase and identifying software flaws. The company noted that similar capabilities exist in other publicly available models ^[1].

Anthropic characterized the government action as a misunderstanding and expressed its intention to work towards restoring access as soon as possible ^[1]. The company had launched Fable 5 publicly on June 9, 2026, as a version of its Mythos-class model with guardrails that defer high-risk queries to a less capable model, Claude Opus 4.8 ^[5].

What Is Known

The U.S. government has issued an export control directive requiring Anthropic to suspend access to its advanced AI models for foreign nationals due to national security concerns ^[1][3]. Anthropic has disabled access to the Fable 5 and Mythos 5 models globally, citing the infeasibility of selective enforcement ^[1]. Amazon’s report to the White House about a potential jailbreak vulnerability in the Mythos model prompted the government’s directive ^[4].

What Remains Unclear

The severity of the alleged jailbreak vulnerability remains disputed. While Anthropic and some experts argue it was narrow and minor, government sources and Amazon’s report suggest it posed a significant national security risk ^[1]. Additionally, the timeline for restoring access to the models is not specified ^[1].