Anthropic restored access to its Mythos 5 AI model for select U.S. critical infrastructure groups after a U.S. government review. The company announced on X that it's redeploying Mythos 5 to organizations defending national assets, though exact numbers remain undisclosed. Semafor reported over 100 institutions now have access, including major corporations and agencies. The move follows a June 12 suspension triggered by national security concerns, including fears that a China-linked group had infiltrated the tool. The U.S. Commerce Department approved the redeployment after Anthropic implemented safeguards, per a letter from Commerce Secretary Howard Lutnick.

The blockade stemmed from worries Mythos 5 could be exploited by malicious actors, with Amazon and others warning of vulnerabilities. The government feared foreign nationals, even U.S.-based ones, could misuse the AI. Anthropic halted access to Mythos 5 and its public-facing Fable 5 model, citing compliance with national security directives. While Mythos 5 access was restored swiftly, Fable 5 remains under review, with no timeline for reinstatement. The company emphasized ongoing collaboration with authorities to expand access while maintaining strict controls.

Lutnick's letter confirmed Anthropic's daily negotiations with the government over the past two weeks led to today's decision. The redeployment prioritizes trusted partners operating critical infrastructure, though the full scope of institutions involved isn't public. This marks a pivotal shift from the initial suspension, balancing security concerns with the need for advanced cybersecurity tools.

The restoration of Mythos 5 underscores the U.S. government's cautious approach to AI deployment, requiring companies like Anthropic to meet rigorous standards. While the tool's return aids critical defense efforts, the unresolved status of Fable 5 highlights lingering concerns about broader AI accessibility. Anthropic's ability to navigate these rules may shape future tech policy frameworks.