EVERYTHING
According to WIRED, US Commerce Secretary Howard Lutnick has authorized Anthropic to redeploy its most powerful model, Mythos 5, to a select group of trusted partners. This includes large corporations and government agencies that have been vetted to ensure they meet specific safety standards. The decision follows a period of tension where the company had completely disabled access to the models due to federal export control directives.
Security safeguards and regulatory progress
In a letter addressed to Anthropic cofounder and chief compute officer Tom Brown, Lutnick stated that he determined appropriate safeguards are in place to mitigate the risks associated with the model. The government’s stance reflects a shift from an initial total block toward a managed rollout for specific use cases. However, the administration has not yet approved a broad public release of the technology.
The restoration is specifically targeted at cyber defenders and infrastructure providers who require high-level capabilities to protect digital systems. Anthropic spokesperson Eduardo Maia Silva confirmed that the company is currently working to provision these approved providers as quickly as possible. Key details regarding the current status include:
Context of the regulatory standoff
The friction began roughly two weeks ago when the White House issued an export control directive requiring Anthropic to limit access for foreign nationals, including those living in the United States. This prompted Anthropic to shut down the models entirely. The administration's concerns were heightened after reports surfaced regarding a South Korean telecommunications firm with potential ties to China receiving access. Additionally, Amazon and the National Security Agency expressed fears that Fable 5 could be jailbroken.
While Mythos 5 is returning to a limited user base, Anthropic continues to lobby for the restoration of its consumer-facing model, Fable 5. The company has sent senior members from its cybersecurity and AI safety teams to Washington, DC, to negotiate with officials. This ongoing dialogue highlights the broader challenge of establishing a permanent policy framework for high-stakes AI releases in an era of increasing geopolitical tension.