Restricted access granted to trusted partners
Two weeks after the Trump administration imposed a ban that forced Anthropic to pull its advanced cybersecurity models, Mythos 5 and Fable 5, from the market, the government has softened its stance. According to reports from Semafor and Reuters, the Commerce Department has now authorized Anthropic to make Mythos 5 available to more than 100 specific U.S. government agencies and companies. This authorization notably extends to non-American employees within those organizations, including Anthropic's own non-American staff, who were originally barred from accessing the models under the ban. Commerce Secretary Howard Lutnick communicated the decision in a letter to Anthropic's chief compute officer Tom Brown on Friday, stating that appropriate safeguards were in place to permit certain trusted partners access to the Claude Mythos 5 Model. The directive, however, did not address the release of Fable 5, a version of Mythos 5 that was widely released days before the ban due to its purportedly stronger protections.
Background of the ban and safety concerns
The initial ban, implemented on June 12, 2026, came after security researchers allegedly bypassed the safety guardrails on both Mythos 5 and Fable 5 with relative ease. Anthropic had marketed these models as powerful cybersecurity tools, but the vulnerabilities raised significant concerns within the Trump administration about potential misuse by unauthorized parties. The ban prohibited any non-American individuals from accessing the models, effectively halting their deployment across many organizations with international workforces. Anthropic complied by pulling both models from the market, but the company continued to work closely with government officials to address the security issues. The Fable 5 model, which had been released more widely just days before the ban, was said to incorporate additional protections, though these proved insufficient against determined researchers. This incident highlighted ongoing tensions between rapid AI deployment and national security considerations.
Anthropic's response and path forward
Anthropic publicly acknowledged the development in a post on X on Friday, stating that since June 12, the company had been working closely with the U.S. government to restore access to both Claude Mythos 5 and Fable 5. The company confirmed that the government had notified them that Mythos 5 could be redeployed to a set of U.S. organizations that operate and defend critical infrastructure. Anthropic emphasized that it was restoring access for these organizations quickly and continuing to work with the government to expand access to Mythos 5 and make Fable 5 available for general use again. The company did not immediately respond to requests for comment from TechCrunch, but the public statement signaled a cooperative stance with the administration. This partial restoration of access represents a significant step in resolving the conflict, though the future of Fable 5 remains uncertain.
Implications for AI regulation and industry
This episode marks a notable instance of direct government intervention in the deployment of advanced AI models, setting a precedent for how future security concerns might be handled. The Trump administration's decision to allow access to over 100 entities, including those with non-American employees, suggests a calibrated approach that balances national security with the operational needs of critical infrastructure organizations. For the AI industry, this case underscores the importance of robust safety guardrails and the potential consequences when they fail. Anthropic's experience may influence how other AI labs design and test their models before release, particularly for applications in sensitive sectors like cybersecurity. The partial nature of the restoration—leaving Fable 5 in limbo—also indicates that the government is taking a measured, case-by-case approach rather than issuing blanket approvals. This could lead to more structured frameworks for AI model certification and access control in the future.
Competitive landscape and market impact
The temporary ban and subsequent partial restoration of Mythos 5 have affected Anthropic's competitive standing in the AI market, particularly as Asian startups have reportedly launched similar models during the export ban. This development highlights the global nature of AI development and the challenges of unilateral restrictions. Anthropic's Claude brand has been gaining traction among paid consumers, challenging OpenAI's ChatGPT in the market. However, the uncertainty around model availability could prompt some organizations to explore alternatives. The administration's decision to allow access for critical infrastructure operators may help mitigate immediate disruptions, but the long-term impact on Anthropic's business relationships remains to be seen. The company's ability to quickly restore access and continue negotiations for broader deployment will be crucial in maintaining its market position. This incident also raises questions about the security and reliability of AI models in high-stakes environments.