Anthropic shuts down Claude Fable 5 and Mythos 5 days after launch: Here’s why

AI company Anthropic has been asked by the US government to disable access to two of its latest AI models, Claude Fable 5 and Claude Mythos 5, following concerns linked to national security and cybersecurity capabilities. Also Read: ChatGPT prices may drop as OpenAI looks to stay ahead of Anthropic

The directive was reportedly issued on June 12, and Anthropic says it has already complied. The order affects access to both models, while the company’s other AI models remain available. Also Read: Asked ChatGPT for a number between 1 and 100? Here’s why 73 often shows up

The move has sparked debate across the AI industry, especially because Fable 5 had only launched days earlier and Mythos 5 was never widely released to the public. Also Read: Anthropic launches Claude Fable 5: Public version of its most powerful AI model arrives with safeguards

What are Claude Fable 5 and Mythos 5?

Claude Mythos 5 was Anthropic’s most advanced AI model focused on cybersecurity. According to the company, the model showed an unusual ability to identify software vulnerabilities across operating systems and web browsers.

Instead of releasing it publicly, Anthropic kept the model under a restricted program called Project Glasswing. Access was reportedly limited to around 50 vetted organisations, including major technology companies such as Amazon, Apple, Google, Microsoft, and cybersecurity firm CrowdStrike.

Fable 5, on the other hand, was introduced as a public-facing version of Mythos 5. Anthropic said it included additional safeguards designed to block sensitive responses in areas such as cybersecurity and biology while still offering strong general-purpose AI capabilities.

What is the issue?

According to Anthropic, the government’s concerns appear to be tied to reports that Fable 5’s safeguards could be bypassed through a jailbreak technique.

A jailbreak is a method used to make an AI model ignore certain restrictions or safety measures. Anthropic says the evidence shared with it involved a limited scenario where the model could analyse code and identify software vulnerabilities.

The company argues that similar capabilities already exist in several publicly available AI models and are commonly used by security researchers and software teams to identify weaknesses before attackers can exploit them.

The US government, citing national security authorities, has issued an export control directive to suspend all access to Fable 5 and Mythos 5 by any foreign national, whether inside or outside the United States, including foreign national Anthropic employees.

The net effect of…

— Anthropic (@AnthropicAI) June 13, 2026

Anthropic also says that its most important safety protections operate through separate systems that continue to monitor and block dangerous outputs even if a user manages to push past an initial refusal.

Why the government stepped in

The concern appears to be less about everyday AI usage and more about advanced cybersecurity capabilities.

Mythos 5 was designed to find software weaknesses at a high level. While that can help organisations improve security, the same capability could potentially be misused if it falls into the wrong hands.

As a result, officials reportedly decided to take a cautious approach and ordered access to both models to be suspended while concerns are reviewed.

Anthropic disagrees with the decision

Anthropic has publicly pushed back against the order and says a limited jailbreak should not be enough reason to remove a model that had already been released.

The company argues that applying the same standard across the industry could make it much harder for AI companies to launch new models in the future.

Add Techlusive as a Preferred Source

The development comes at an interesting time for Anthropic, which has built much of its reputation around AI safety. The company has often positioned itself as one of the more cautious players in the AI race, but that same focus on highlighting potential risks around advanced models may now be attracting additional scrutiny from regulators.