Why has the US lifted restrictions on Anthropic’s AI models Fable 5 and Mythos 5?

The US government and Anthropic on Tuesday reached an agreement to restore access to the company’s latest artificial-intelligence (AI) models, Fable 5 and Mythos 5.

Anthropic, in a statement, announced that it will soon begin restoring access globally to its most powerful AI models.

Fable 5, which is a part of Anthropic’s Claude family of AI systems, will again be available to global users on the Claude platform, Claude.AI, and Claude Code starting Wednesday.

“We’re grateful to our users for their patience, and to everyone who worked with us on redeploying the models,” Anthropic said in a post on X.

The US Commerce Secretary Howard Lutnick, in a letter written to Anthropic informing about the decision, said that Anthropic had agreed to work diligently with the US government ​on protocols for Mythos, Fable and future models and to inform the US government of any malicious activity.

Why was Fable 5 and Mythos 5 banned?

Anthropic had launched Fable 5 and Mythos 5, the most powerful systems it has ever publicly released, on June 9.

Fable 5 included guardrails that prevented the AI system from assisting with a wide range of cybersecurity-related tasks. The company said measures were in place to prevent malicious cyber purposes.

However, on June 12, Anthropic was forced by the US government to take both models offline citing a potential risk to national security.

The company said it was told to suspend all access “by any foreign national, whether inside or outside the United States, including foreign national Anthropic employees.”

Senior US administration officials became worried that users might be able to circumvent Fable 5’s guardrails.

Why has the ban been lifted now?

At the time of the ban, Anthropic had released a statement saying that the “minor” vulnerabilities mentioned by the US government as the grounds of restrictions imposed on the AI models “all appear relatively simple”. More importantly, it said: “We have found that other publicly-available models are able to discover them as well without requiring a bypass.”

Experts agreed. It was widely discussed that if Fable and Mythos were blocked on the grounds cited by the government, then competitor models would have to be blocked too.

Also, the government’s crackdown on Anthropic coincided with a swift rise in Chinese open-source models. These models are not just equally capable, but also cheaper than some of the most powerful US models. Hence, investors raised concerns that a ban on Anthropic could turn out to be a boon for these Chinese developers.

More importantly, the decision to impose restrictions on AI models could have set a precedent that the US government needs to approve every frontier model release.

Its effect was visible. ChatGPT creator OpenAI announced last week that it would launch its latest AI model series, GPT-5.6, for only a “small group of trusted partners” at first, following pressure by the US government to stagger the release.

Earlier in May, the National Institute of Standards and Technology announced that Google, Microsoft and xAI will share unreleased versions of their AI models with the government to curb cybersecurity threats.

Stakeholders are still assessing how governments can address risks from the most powerful AI models. Until then, whether government reviews of AI releases should become a long-term default and if it is justified remains to be seen.