The Commerce Department has cleared the Anthropic Mythos 5 cybersecurity model for redeployment, lifting a two-week block that had cut off the company’s most capable security tool. In a letter dated Friday, Commerce Secretary Howard Lutnick said the model can return to roughly 100 U.S. organizations that operate and defend critical infrastructure.
“I have determined that appropriate safeguards are in place to permit certain trusted partners to access the Claude Mythos 5 Model,” Lutnick wrote, according to Semafor, which first reported the decision. Anthropic confirmed the move in a public statement on Friday.
https://x.com/AnthropicAI/status/2070665903440871779
Why Mythos 5 was pulled in the first place
Anthropic disabled customer access to both Mythos 5 and the public-facing Fable 5 on June 12, complying with a government order to suspend use by foreign nationals. We covered that abrupt suspension when it landed.
The restrictions followed warnings from Amazon and other firms about possible jailbreaking. Regulators were also concerned that a China-linked group had gained access. Semafor reported that the access ran through a South Korean telecommunications provider.
Mythos is Anthropic’s strongest cyber model, built to find and exploit software flaws at speed. That capability is what made officials nervous in the first place. For a primer on how the two models differ, see our explainer on Fable 5 and Mythos 5.
What the clearance covers, and what is still on hold
The redeployment is narrow by design. Access returns only to vetted government agencies and private companies, and only for defensive cyber work. Lutnick addressed his letter to Tom Brown, Anthropic’s chief compute officer, and said the company has “committed to work with the U.S. government on protocols and standards and releases.” He credited two weeks of near-daily talks for ending the standoff.
A Commerce spokesman, Benno Kass, framed the outcome as a balance. “In just two weeks, we have worked diligently to ensure America remains the global leader in AI while safeguarding our security,” he said.
Fable 5 stays under review. People close to the talks told Semafor that a release is likely, though the timeline is unclear. The decision mirrors a parallel arrangement for OpenAI, which is releasing ChatGPT 5.6 to government-approved customers first.
Anthropic said it would keep working with officials to widen access over time.
