Trump Administration Allows Anthropic's Mythos 5 for Critical US Entities

The Trump administration has significantly softened its stance on Anthropic's powerful cybersecurity-oriented models, specifically allowing its Mythos 5 model to be deployed to over 100 U.S. government agencies and companies. This decision comes just two weeks after a sweeping ban forced Anthropic to pull both Mythos 5 and its counterpart, Fable 5, from the market, creating uncertainty around the use of advanced AI in sensitive sectors. The initial ban had notably restricted access for non-American employees, a point of contention that has now been addressed in the new directive. Under the updated policy, Anthropic is now permitted to make Mythos 5 available to a broad spectrum of specific U.S. organizations, including critical infrastructure operators and defense entities. Crucially, the new directive explicitly allows non-American employees within these organizations, as well as Anthropic's own non-American staff, to access the model. This reversal marks a significant shift from the original ban, which had broadly prohibited non-U.S. citizens from interacting with these advanced AI systems due to national security concerns. Commerce Secretary Howard Lutnick confirmed the administration's revised position in a missive to Anthropic’s chief compute officer Tom Brown. According to Semafor, Lutnick stated, "I have determined that appropriate safeguards are in place to permit certain trusted partners to access the Claude Mythos 5 Model." This suggests that the government and Anthropic have collaborated to establish robust security protocols that address the administration's initial concerns, paving the way for the model's redeployment in vital U.S. operations. While Mythos 5 gains a path back to deployment, the administration's directive did not address the release of Fable 5. This model, a version of Mythos 5, was widely released just days before the initial ban, marketed with enhanced protections. However, both models were subsequently pulled after security researchers reportedly managed to easily bypass Fable 5's supposed guardrails, raising questions about their inherent security. Anthropic has not yet commented on the specifics of Fable 5's future in this immediate context. Anthropic publicly acknowledged the progress on Friday via a post on X, confirming the collaborative efforts with the U.S. government. The company stated, "Since June 12, we’ve been working closely with the US government to restore access to Claude Mythos 5 and Fable 5. Today, the government notified us that Mythos 5, our strongest cybersecurity model, can be redeployed to a set of US organizations that operate and defend critical infrastructure." Anthropic also indicated that it is actively continuing discussions with the government to further expand access to Mythos 5 and to eventually make Fable 5 available for general use again, signaling a phased approach to re-integration. This measured re-introduction of Mythos 5 underscores the complex balance between fostering technological innovation and ensuring national security, particularly in the rapidly evolving field of AI. The move allows critical U.S. entities to leverage advanced AI for cybersecurity, while also setting a precedent for how governments and AI developers can work together to establish trust and implement necessary safeguards for powerful, dual-use technologies.