US Government Partially Lifts Restrictions on Anthropic's Mythos 5 AI Model

US Government Partially Lifts Restrictions on Anthropic’s Mythos 5 AI Model The U.S. government has partially reversed its abrupt shutdown of Anthropic’s most powerful AI cybersecurity system, Mythos 5, reopening it to a tightly controlled group of users while keeping its consumer counterpart, Fable 5, offline.

From sudden shutdown to two‑week freeze

On June 12, the Commerce Department issued an order that effectively forced Anthropic to pull both Mythos 5 and the more widely accessible Fable 5 from the market, stunning the AI industry and disrupting cyber operations that had begun to rely on the tools. The move was framed as an export‑control step meant to curb risks from the models’ advanced cyber capabilities, initially barring access by foreign nationals, including Anthropic’s own non‑U.S. employees.

Over the next two weeks, Anthropic entered what one report described as a “rollercoaster negotiation process” with the Trump administration over how and when Mythos 5 might safely return.

June 26: Limited access restored

On June 26, Commerce Secretary Howard Lutnick informed Anthropic that license requirements for Mythos 5 were being revised, permitting exports and in‑country transfers to a defined set of entities listed in an annex and to their foreign national staff, as well as to Anthropic’s own non‑U.S. employees. Lutnick wrote that appropriate safeguards were now in place and that Anthropic’s engagement with the government had “yielded significant progress.”

Reports say more than 100 U.S. government agencies and companies, including critical‑infrastructure providers, are now authorized to use Mythos 5, with access also extended to non‑American employees at those organizations.

Anthropic confirmed it had “received notice from the US government that Mythos 5, our strongest cybersecurity model, can be redeployed to a small group of cyber defenders and infrastructure providers,” adding that it is working to restore access “as quickly as possible” and continues to seek broader availability and a path to re‑release Fable 5 for general use.

Ongoing uncertainty and future controls

While this carve‑out mirrors a similar arrangement for OpenAI’s GPT‑5.6, export controls remain in place for all unapproved organizations, and Fable 5 is still in limbo with no public timeline. Lutnick explicitly reserved the right to tighten or expand the rules and to change the list of approved entities “at any time,” underscoring that AI export policy for high‑capability models like Mythos remains ad hoc and subject to rapid change.

Continue reading https://foxvector.com/stories/019f078a-1586-12af-72d2-1fdf6f6fef03