Anthropic has reinstated Claude Fable 5, but users will face a temporary usage restriction.

      Anthropic has obtained approval from the US government to reinstate its AI Model, just weeks after a security researcher discovered a method to bypass its safeguards, resulting in the model's shutdown.

      Anthropic will be reinstating full access to Claude Fable 5 starting tomorrow, weeks after a US government order compelled the company to suspend the model for all users. This directive was issued on June 12 and required Anthropic to prevent foreign nationals from using Fable 5 and the more advanced Mythos 5 model. Due to the immediate enforcement of this rule and Anthropic's inability to verify a user's nationality in real-time, the company decided to suspend both models entirely instead of risking a violation.

      What led to the shutdown

      These restrictions followed a report from Amazon researchers, who discovered a way to induce Fable 5 to reveal software vulnerabilities that its safeguards were meant to protect against. In one case, the model generated code that illustrated how a vulnerability could be exploited.

      We’ve received notification that the Department of Commerce has removed export restrictions on Claude Fable 5 and Mythos 5. We will start restoring access tomorrow and will provide an update shortly. We appreciate our users' patience and everyone who assisted us on…— Anthropic (@AnthropicAI) June 30, 2026

      Anthropic later confirmed that several less advanced models, such as Opus 4.8 and other competitors, were capable of yielding similar outputs, indicating that the bypass was not solely dependent on unique capabilities of Fable 5.

      Terms for return and new protections

      Fable 5 will be accessible globally starting July 1 for users on the Claude Platform, Claude.ai, Claude Code, and Claude Cowork. For Pro, Max, Team, and select Enterprise plans, Fable 5 usage will count towards up to 50 percent of the weekly usage limit until July 7. After this date, the model will utilize usage credits.

      Claude Fable 5 will be globally available again tomorrow. Following a series of constructive discussions with the US government, we are redeploying the model with a new set of classifiers designed to target and restrict more cybersecurity tasks. In the immediate term, some routine activities like coding…— Anthropic (@AnthropicAI) July 1, 2026

      Anthropic reports that it has developed a new safety classifier in partnership with the government, which successfully blocks the identified technique in over 99 percent of instances. If a request to Fable 5 is blocked, users will be notified, and the query will automatically be redirected to Opus 4.8.

      This clearance comes after a similar government review that reinstated access to Mythos 5 on June 26 for a limited number of approved US organizations. Anthropic states it is collaborating with the government to extend access to more domestic and international partners. Additionally, the company is working with Amazon, Microsoft, and Google on establishing a shared industry standard for assessing the severity of AI jailbreaks moving forward.