US Allows Limited Access to Anthropic’s Mythos 5 AI Model for Select Partners

In June, the Trump administration imposed export controls on Anthropic’s frontier AI models Mythos 5 and its lighter‑weight counterpart Fable 5, citing national‑security risks such as potential jailbreaks that could enable malicious actors to weaponize the technology. After weeks of negotiations, the U.S. Commerce Department, led by Secretary Howard Lutnick, issued a letter stating, “I have determined that appropriate safeguards are in place to permit certain trusted partners to access the Claude Mythos 5 Model.” The partial lift authorises the model’s use by roughly 100 vetted U.S. organisations involved in cyber‑defence and critical‑infrastructure protection, while Fable 5 remains blocked. At the same time, OpenAI has agreed to a staggered release of its newest system, GPT‑5.6, limiting the rollout to a small list of government‑approved partners. The actions reflect a broader effort by the Trump administration to tighten oversight of high‑risk AI, addressing concerns about rapid vulnerability discovery, disinformation, and the potential for AI‑driven cyber‑attacks. Anthropic has said it will continue working with officials to expand access and to restore Fable 5, underscoring the emerging balance between AI innovation and security regulation.