US Government Narrows Access to Advanced AI Models Amid Security Concerns

The U.S. Commerce Department has partially lifted an export‑control ban on Anthropic’s high‑performance AI model Claude Mythos 5, allowing more than 100 trusted organizations—including federal agencies and large enterprises—to resume access, while the related model Fable 5 remains fully blocked. The partial release follows a two‑week shutdown triggered by national‑security worries that the models could be used to discover software vulnerabilities quickly.

At the same time, OpenAI has restricted the preview version of its GPT‑5.6 series to a small, vetted group of partners after a request from the U.S. government, citing the same fear of cyber‑weaponisation. Separate from these releases, AI‑startup Legion has filed a lawsuit in Washington, D.C., claiming that the government‑mandated removal of Anthropic’s “Fable 5” and “Mythos 5” models from its product devastated its business, arguing the action was “existential” for its operations.

These actions illustrate a tightening regulatory stance in the United States toward frontier AI systems, with export‑control measures and selective access intended to mitigate perceived risks of misuse while still permitting limited, controlled deployment for trusted users.