US Eases Restrictions on Anthropic's Claude Mythos 5 for Trusted Partners

The United States government has announced a partial lifting of its ban on Anthropic's advanced artificial intelligence model, Claude Mythos 5, two weeks after imposing stringent export controls. This significant development allows more than 100 "trusted" US companies and institutions, including numerous Fortune 500 entities, to regain access to the powerful AI system. The initial ban, enacted on June 12, 2026, had forced Anthropic to abruptly disable access to both Mythos 5 and its companion model, Fable 5, for all users, including foreign nationals. This drastic measure was taken amidst escalating national security concerns that these frontier models could be exploited by adversarial nations like China and Russia for military or intelligence purposes. US Commerce Secretary Howard Lutnick confirmed the decision in a letter to Anthropic, explicitly stating that "appropriate safeguards are in place to permit certain trusted partners to access the Claude Mythos 5 Model." This partial reversal is the culmination of intensive, daily discussions between Anthropic and the Commerce Department, aimed at thoroughly addressing the government's security apprehensions. Anthropic has expressed considerable satisfaction with the progress made, noting that Mythos 5, which is particularly recognized for its robust cybersecurity capabilities, can now be redeployed to organizations critical for defending national infrastructure. The company is actively engaged in restoring access for these newly approved providers and continues its dialogue with the government to further expand availability for Mythos 5 and eventually reintroduce Fable 5 for general use, which is reportedly on track to return soon. The government's initial aggressive stance on AI model oversight, driven by fears of potential misuse and "jailbreaking" capabilities, also impacted rival AI developers. OpenAI, for instance, had to delay the full public launch of its GPT-5.6 model, limiting its access to a select group of vetted partners whose details were shared with authorities. This situation underscores the growing tension between fostering rapid AI innovation and implementing robust national security safeguards. While the current directive permits non-US citizens employed by approved companies to utilize Mythos 5, the broader restrictions on other entities and the continued limitations on the Fable 5 model highlight the dynamic and evolving regulatory landscape for advanced AI technologies. This development marks a pivotal moment in the ongoing dialogue between leading AI developers and governmental bodies concerning the responsible and secure deployment of cutting-edge artificial intelligence. Anthropic had previously objected to the broad export controls, arguing that such standards could "essentially halt all new model deployments for all frontier model providers" if applied across the industry. The company also noted that "perfect jailbreak resistance is not currently possible for any model provider." This partial lift represents a compromise, allowing critical US entities to leverage advanced AI while the government maintains a cautious approach to broader distribution.