Anthropic disabled worldwide access to its Claude Fable 5 and Claude Mythos 5 AI models on June 12, 2026 [1].

The shutdown marks a rare instance of the U.S. government intervening to pull a commercial AI product from the global market. This action highlights the growing tension between rapid AI deployment and national security oversight as models become more powerful.

The decision followed a directive from the U.S. government citing concerns over a potential AI "jailbreak" [4]. A jailbreak occurs when a user bypasses the safety filters of a model to force it to generate prohibited or dangerous content. While the government viewed this vulnerability as a critical risk, Anthropic disputed the necessity of the recall.

In a statement on the company blog, Anthropic said, "We disagree that the finding of a narrow potential jailbreak should be cause for recalling a commercial model deployed to hundreds of millions of people" [2].

The models were taken offline just days after their initial launch [3]. While some reports suggest a censored version of Fable 5 remains available, other reports indicate that both the Fable and Mythos models were disabled for all users worldwide [5].

The company's pushback suggests a conflict regarding the threshold of risk that justifies a total shutdown. By citing the impact on hundreds of millions of users [2], Anthropic said the scale of the disruption was caused by the government's security mandate.

“Anthropic disabled worldwide access to its Claude Fable 5 and Claude Mythos 5 AI models”

This incident establishes a precedent for government-mandated recalls of AI software based on theoretical security vulnerabilities. It suggests that the U.S. government may prioritize the prevention of 'jailbreaks' over the commercial availability of high-capability models, potentially leading to more stringent pre-release audits for AI startups.