US Allows Limited Release of Anthropic’s AI Model Mythos 5
6 verified3 unconfirmed1 contested
The Trump administration has eased restrictions on Anthropic’s advanced AI model Claude Mythos 5, permitting access for more than 100 US organizations including large corporations and government agencies. Commerce Secretary Howard Lutnick informed Anthropic co-founder Tom Brown in a letter dated June 26 that the government found “appropriate safeguards are in place” after the company worked to address risks. However, the export control directive from June 12 remains largely in effect, and the consumer-facing version Claude Fable 5 is still restricted with no announced timeline. Anthropic is continuing discussions with the White House to expand access. The partial reinstatement follows two weeks of negotiations after the administration initially barred foreign nationals from accessing the models, prompting Anthropic to disable them entirely.
What’s verified
The Trump administration allowed Anthropic to restore access to Mythos 5 for a select group of US organizations, including cyber defenders and infrastructure providers.
Commerce Secretary Howard Lutnick sent a letter to Anthropic co-founder Tom Brown stating that “appropriate safeguards are in place” after Anthropic worked with the US government on risk mitigation.
The export control directive issued on June 12 remains in effect, except for the exception that permits approved organizations and Anthropic to allow foreign national employees to access Mythos 5.
Claude Fable 5, the public-facing version of Mythos 5, remains restricted with no clear timeline for release.
Anthropic is in discussions with the White House about restoring Fable 5, and both parties hope the resolution will inform a policy framework for future model releases.
OpenAI announced it was delaying the release of its GPT-5.6 models in response to a request from the Trump administration, with OpenAI stating it does not believe the limited government access process should become the long-term default.
Where accounts differ
One source identifies Anthropic’s spokesperson as Eduardo Maia Silva, while another identifies the spokesperson as Danielle Ghiglieri. The content of the statement attributed to the spokesperson is consistent across both sources.
Not yet confirmed
The specific number of organizations approved to access Mythos 5 is reported by only one source as “more than 100.”
The exact timeline for a potential resolution on Fable 5 is not specified in either source.
It is unclear whether the National Security Agency’s concerns about Fable 5 being jailbroken, reported by one source, directly influenced the administration’s decision.
Key figures
Tom Brown – Anthropic co-founder and chief compute officer
Howard Lutnick – US Commerce Secretary
Eduardo Maia Silva / Danielle Ghiglieri – Anthropic spokespersons (names differ per source)
Dean Ball – head of strategic futures team at OpenAI and former White House AI adviser
Sarah Heck – Anthropic's public policy chief
