The US government granted Anthropic permission on June 26, 2026 to release its Claude Mythos 5 model to roughly 100 companies and federal agencies. Commerce Secretary Howard Lutnick confirmed in a letter that “appropriate safeguards are in place to permit certain trusted partners to access the Claude Mythos 5 Model,” ending a two-week block driven by national security concerns.
The decision marks a careful middle path between open release and outright restriction for one of the most capable AI models yet built. It follows a period of intense debate over how to govern frontier AI, a theme we explored in our coverage of the accelerating AI model race.
Why the Government Blocked Mythos 5 First
Two weeks before the approval, the administration imposed export controls on Mythos, citing national security authorities. Anthropic disabled access to both Mythos 5 and its cousin Fable 5 to comply, after warnings from Amazon and other partners that the models could be jailbroken for malicious purposes.
The concern centered on the models’ advanced cybersecurity capabilities, which officials feared could empower attackers if released broadly. That worry mirrors the scrutiny OpenAI faced, as covered in our reporting on AI competition and safety debates.
Who Gets Access to Claude Mythos 5
More than 100 companies and institutions, including many Fortune 500 firms, will now be able to use Mythos 5. As CNBC reported, access is limited to vetted cyber defenders and critical infrastructure providers rather than the general public.
This selective rollout reflects a new governance model where the most powerful AI systems reach trusted organizations under government oversight rather than launching openly to everyone at once. It effectively creates a tiered access system for frontier AI.
Industry observers say the approach could become a blueprint for how Washington handles future frontier releases, balancing economic competitiveness against the risk that advanced models fall into the wrong hands. Other AI developers are watching the precedent closely.
Mythos 5 as a Cyber-Defense Tool
Anthropic describes Mythos 5 as its strongest cybersecurity model. The same capabilities that worried regulators also make it valuable for defense, allowing security teams to detect vulnerabilities and respond to threats faster than human analysts working alone.
By restricting access to defenders and infrastructure operators, the government is betting that Mythos 5 will strengthen national security more than it endangers it. As Bloomberg noted, the trusted-partner framework is designed to capture the upside while limiting the risk.
What the Letter Left Out: Fable 5
Notably, the government’s authorization did not extend to Fable 5, the less powerful sibling model that was also disabled during the export-control episode. Fable 5 remains restricted, leaving Anthropic with a partial resolution rather than a full return to normal operations.
The omission suggests regulators are evaluating each model individually rather than lifting controls wholesale, a sign that AI export policy will remain granular and case-by-case. For readers weighing how safe these systems are, see our explainer on Anthropic’s privacy and data policy.
Related Articles
The AI Model Race Heats Up in June 2026
Trust Post Desk
A journalist and editor at TrustPost.org covering world and national news, technology updates and human-interest stories. They check every fact, interview sources in person or online, and aim to deliver clear, accurate reporting. Their work ranges from breaking news to in-depth features and daily newsletters. Outside the newsroom, they follow emerging trends and engage with readers on social media.
