Anthropic will resume access to its latest artificial intelligence (AI) model, Fable 5, on July 1 (local time).
On June 30 (local time), the company announced that it had reached an agreement with the U.S. government to restore access to Fable 5. The breakthrough came after Anthropic agreed to address concerns related to security safeguards, including allegations raised by Amazon researchers that the publicly available version of the Claude Mythos model could be exploited for cyberattacks and that Fable 5's security protections could be bypassed.
U.S. Secretary of Commerce Howard Lutnick said, “Over the past two weeks, the government has worked closely with Anthropic to evaluate Fable 5, strengthen cooperation across the board, and reinforce U.S. leadership in AI.” Tom Brown, Anthropic's Chief Computing Officer (CCO), also emphasized during discussions with government agencies that the Fable 5 model is safe for public use.
Following the agreement, Anthropic confirmed that access to Fable 5 would be restored after approximately 20 days of suspended service.
The move comes as the U.S. government increases oversight of next-generation AI model releases by domestic AI companies. President Donald Trump recently signed an executive order outlining a new approach that requests companies to provide the federal government with access to new AI models 30 days before their public release.
White House Chief of Staff Susie Wiles said, “The government and the private sector have cooperated in unprecedented ways, establishing an unprecedented 'America First' foundation. Our shared top priority is to deploy the best technology as quickly and safely as possible.”
The U.S. government indicated that lifting the export restrictions on Fable 5 was made possible after Anthropic addressed vulnerabilities identified by organizations including Amazon and established procedures to prevent similar issues in the future. Officials also signaled that improving communication with AI developers will be a key objective going forward. Meanwhile, limited access within U.S. companies and government institutions to Mythos 5, which remains restricted for foreign users, was partially restored last week.
Previously, on June 12, the U.S. government prohibited all foreign nationals from using Anthropic's Mythos 5 and Fable 5 models over safety concerns. The restrictions also applied to foreign nationals working in the United States and even Anthropic researchers, prompting the company to suspend access to its latest AI models for all users.
Separately, Anthropic unveiled its new agentic AI model, Claude Sonnet 5, on the same day. The model features enhanced autonomous capabilities, enabling it to formulate plans independently and use tools such as web browsers and terminals to complete tasks with minimal human intervention.
According to Anthropic, Sonnet 5 delivers improved performance over its predecessor, Claude Sonnet 4.6, across key agentic AI capabilities, including reasoning, tool use, coding, and knowledge work. The company says it offers performance approaching that of its flagship Claude Opus 4.8 model at a lower price point.
The model is also designed to complete complex tasks more reliably. Early users reported that Sonnet 5 successfully finished tasks that previous models abandoned midway and even verified its own results without additional prompting. Anthropic added that the model demonstrates stronger resistance to malicious requests and prompt injection attacks while reducing hallucination rates and overly agreeable responses.
