The US government has lifted restrictions on Anthropic's most powerful AI models, Claude Fable 5 and Mythos 5, allowing the company to restore global access. Commerce Secretary Howard Lutnick communicated this decision to Anthropic in a letter, stating that the Trump administration was satisfied with Anthropic's efforts to address the risks associated with these models. This reversal comes roughly three weeks after the government abruptly forced Anthropic to halt access to these models on June 12, citing national security concerns and vulnerabilities.
The initial restrictions were imposed after researchers reportedly found a way to bypass Fable 5's safeguards, accessing the more capable cybersecurity system, Mythos 5, which operates underneath it. CIA Director John Ratcliffe had previously compared the capabilities of advanced AI models to "digital nuclear weapons," seemingly in defense of the administration's hardline stance on controlling such technology. Despite Anthropic disputing the severity of the vulnerability, the government's intervention led to the temporary shutdown.
Anthropic has committed to working with Amazon, Microsoft, and Google to develop a standardized framework for assessing and responding to critical vulnerabilities in advanced AI models. While Commerce Secretary Lutnick noted that an export license would no longer be required, Anthropic agreed to proactively detect and address security risks, help develop standards for future models, and report malicious activity to the government. White House Chief of Staff Susie Wiles also expressed gratitude for the cooperation from tech companies in implementing President Trump's executive order on AI and cybersecurity.
Restoration of access will be gradual, with usage capped at up to 50% of normal weekly limits through July 7, before returning to full availability. Anthropic confirmed that Fable 5 would return globally across its products, including the Claude platform, Claude.ai, Claude Code, and Claude Cowork, starting July 1. This development follows a May briefing where the White House informed AI firms like OpenAI, Anthropic, and Reflection AI about a planned executive order that would establish a voluntary framework for reviewing advanced AI models before their release.
