The US Department of Commerce has lifted export controls on the Claude Fable 5 and Claude Mythos 5 models, Anthropic announced on June 30.
We’ve received notice that the Department of Commerce has lifted export controls on Claude Fable 5 and Mythos 5.
We’ll begin restoring access tomorrow, and will share an update soon.
We’re grateful to our users for their patience, and to everyone who worked with us on…
— Anthropic (@AnthropicAI) June 30, 2026
The company will make Fable 5 available to users worldwide starting July 1. The model will be accessible via Claude Platform, Claude.ai, Claude Code, and Claude Cowork. Anthropic stated that access through AWS, Google Cloud, and Microsoft Foundry will be restored “as soon as possible.”
Claude Fable 5 will be available again globally tomorrow.
After a series of productive conversations with the US government, we’re redeploying the model with a new set of classifiers to target and block more cybersecurity tasks. In the near term, some routine tasks like coding…
— Anthropic (@AnthropicAI) July 1, 2026
“After a series of productive discussions with the US government, we are redeploying the model with a new set of classifiers to more accurately identify and block cybersecurity-related tasks. […] We have also begun developing a consensus framework—together with Amazon, Microsoft, Google, and other Glasswing partners—to assess the severity of AI breaches and determine how AI developers should respond,” the post stated.
Access to Mythos 5 has already been restored for some US organizations following government approval on June 26. The company will continue to collaborate with authorities. According to Anthropic representatives, this will include pre-release access to tools and safeguards for evaluation, sharing information on breaches and abuses, and dedicated resources for joint research.
To restore Fable 5, developers have trained a new security classifier designed to restrict behavior described in Amazon’s report to authorities. If a request is blocked, the user will be notified, and the request will be redirected to Claude Opus 4.8. According to the company, the new classifier identifies the specific method in over 99% of cases.
“It is likely impossible to make any AI model completely resistant to jailbreaks,” Anthropic stated.
The company acknowledged that the new measures will lead to more false positives in routine programming and debugging tasks. The team described this as a trade-off for broader access to the model’s other capabilities.
Additionally, Anthropic, along with Amazon, Microsoft, Google, and other Project Glasswing partners, has begun developing a unified approach to assessing the danger of jailbreaks. The company noted that there is currently no industry-wide standard for determining the severity of such security bypasses.
The proposed framework evaluates a jailbreak based on four criteria:
- how much it expands an attacker’s capabilities compared to available tools;
- how many different malicious tasks the bypass covers;
- how easily it can be turned into a real attack;
- how easy it is for others to find or replicate this method.
For the most dangerous cases, the company promises to initiate preliminary protective measures immediately after confirming the threat’s severity. Anthropic is also creating a team for round-the-clock monitoring of channels where jailbreak data is published and launching a HackerOne program for reporting potential security bypasses in Fable 5.
In June, amid restrictions for Anthropic, OpenAI launched the full version of the specialized model for finding, verifying, and fixing vulnerabilities, GPT-5.5-Cyber.
At the end of the month, the company, at the request of US authorities, opened limited access to GPT-5.6 Sol, Terra, and Luna for a small group of trusted partners.
