Anthropic's Mythos 5 returns for select users
What's the story
Anthropic's advanced cybersecurity model, Mythos 5, has been reactivated for a select group of organizations. The move comes after two weeks of intense negotiations with the Trump administration. However, its public-facing counterpart, Fable 5, is still awaiting an agreement for rollout. The US government had previously imposed an export control directive on Anthropic, preventing foreign nationals from accessing either model.
License revision
'Appropriate safeguards are in place'
The US government's letter, dated June 26 and signed by Commerce Secretary Howard Lutnick, said the license requirements were revised after Anthropic "worked with the US government to address risks" associated with both models. The letter also noted that "appropriate safeguards are in place to permit certain trusted partners to access the Claude Mythos 5 Model."
Company statement
We're working to restore access quickly, says Anthropic
In light of the US government's decision, Anthropic spokesperson Danielle Ghiglieri said, "We 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." She added that they are working to provision this approved set of providers and restore their access as quickly as possible.
Control persistence
What about Fable 5?
Despite the revision to license requirements for Mythos 5, the US government has not lifted the export control directive it imposed on Anthropic two weeks ago. This means that foreign nationals, including Anthropic employees, are still barred from accessing Fable 5. However, an exception has been made for Mythos 5 with a select group of organizations approved for access.
Regulatory pressure
Pressure mounts on Trump administration to change approach
As Anthropic's competitors improved their cybersecurity-focused models, pressure has been building on the Trump administration to change its case-by-case regulatory approach. Concerns were also raised within the US AI industry over potential advancements in AI technology by China, while top US AI labs were sidelined. The National Security Agency had also lost access to Mythos 5, further intensifying calls for change.
Strategic alignment
Limited preview for Mythos 5
Anthropic now has a deal similar to rival company OpenAI's: Release the model in a limited preview setting for approved organizations only. These include trusted enterprises and the US government itself. Both AI labs are hopeful for more general availability soon, but it ultimately depends on the Trump administration's decision to approve it.