The US government has lifted export restrictions on Anthropic AI models Fable 5 and Mythos, ending a two-week standoff with the company over concerns that the models could be used by threat actors for malicious purposes.
In a letter sent on June 30, Commerce Secretary Howard Lutnick announced that the Department of Commerce is lifting the restrictions it placed on the company over concerns that the models are susceptible to jailbreaking. The export limitations forced the company to pull its newly released models, Fable 5 and Mythos, from public platforms.
“Anthropic has taken steps in close coordination with the US government to address the risks associated with Claude Mythos 5 and Claude 5. Among other things, Anthropic has agreed to proactively detect and address security risks associated with the models,” Lutnick said in the letter addressed to Anthropic’s Chief Compute Officer, Tom Brown.
The latest development marks the end of two-weeks-long of negotiations between the company and the department. On June 12, the Commerce Department imposed an immediate suspension of the models after Amazon security researchers claimed the AI systems were susceptible to jailbreaking. Anthropic disputed the claim, but withdrew the models immediately.
Following continued talks with Washington, the Commerce Department reversed the hardline restrictions on June 26 to allow the company to sell the models only to its trusted partners.
Anthropic on Tuesday announced that Fable 5 will be available to users across the world from Wednesday on Claude Platform, Claude.ai, Claude Code, and Claude Cowork. “We continue to coordinate with the government to expand access to the broader set of domestic and international partners,” the company said. On Tuesday, the company also released its latest agentic AI, Claude Sonnet 5.
Anthropic’s Advanced AI Framework
The company added that it is also working with Amazon, Microsoft, and Google to develop a shared industry framework to assess the severity of a jailbreak, enabling developers to launch models with “greater safety, and communicate the level of risk consistently to government and industry partners.”
The concerns from the US government come as it continues to assess the disruptive capabilities of the latest AI models. These developments prompted leading companies like OpenAI to release its latest model, GPT 5.6, with limited access.
Security leaders have raised concerns about the negative impact of the Trump administration’s move, with many arguing that hardline limitations prevent AI coders and security teams from using these tools to find and fix flaws in security code “faster than our adversaries.”
They also warned that such limitations put the US on the back foot of the AI race, especially with China. “To pull the best capabilities away from defenders without a good reason when our adversaries are rapidly advancing is dangerous,” they warned.
Washington’s decision to reverse the restrictions also comes in the wake of a recent report that the newly released Chinese open-weight model GLM-5.2, is on par with Mythos in terms of capabilities.
