Anthropic said it regained partial U.S. access to its Mythos 5 AI model after resolving Trump administration concerns about cybersecurity risks, allowing redeployment to a small group of “trusted partners.” Commerce Secretary Howard Lutnick sent a letter to Anthropic’s chief compute officer (seen by Bloomberg News) stating the company addressed risks tied to “Covered Models,” and that Mythos 5 could be released to certain approved organizations. Anthropic said the model could be redeployed to cyber defenders and infrastructure providers. The approval eases a confrontation in which the government abruptly barred global access to Mythos 5 and Fable 5, prompting Anthropic to disable access and intensify negotiations. The letter did not specify changes to restrictions on Fable 5, which Anthropic said remains a continuation of talks. While this is a technology policy decision rather than a campus directive, it affects higher education’s AI research and cybersecurity training ecosystem by shaping what advanced models are accessible for legitimate academic or defensive use under U.S. oversight.