On June 27, 2026, the US Commerce Secretary Howard Lutnick informed Anthropic that its powerful cybersecurity AI model, Claude Mythos 5, could be redeployed to a select group of US organizations operating critical infrastructure. The decision came two weeks after the government ordered Anthropic to block access to Mythos 5 and its consumer variant Fable 5 over national security fears, including potential misuse by foreign adversaries and a reported China-linked group accessing the model. Anthropic had previously suspended all access in response. The partial lift applies to over 100 entities including Fortune 500 companies and government agencies, but Fable 5 remains unavailable pending further talks. Anthropic has committed to daily coordination with the government and expects the incident to inform future AI release policies.
Partial release with criticism of government overreach; includes Altman's concerns
DW English frames the story as a partial release to 'trusted' firms, mentioning the Trump administration's 'thorny relationship' with Anthropic and quoting Sam Altman's criticism of government picking customers. It emphasizes the vetting process and security concerns.
Unrelated article about Italian tax payment system; no coverage of the Anthropic story
This article is entirely unrelated to the topic. It discusses a new Italian service for automatic tax debiting via the F24 model. It does not reference Anthropic, Mythos, or AI regulation.
Wired provides an in-depth account, including the full context of Lutnick's letter, the conditions for foreign national access, and the hope that the incident will inform future AI release policies. It also notes the roles of Amazon and the NSA.
Government grants permission after suspension; focus on quick redeployment and talks for Fable
Engadget reports that the US government gave Anthropic permission to redeploy Mythos 5 to over 100 US institutions, highlighting the company's rapid restoration of access and ongoing work to expand approval, while Fable 5 remains blocked.
NOS reports that Anthropic may now share Mythos with certain government agencies, highlighting the government's fear that the AI tool could be used against them. It also mentions that the government vets access 'customer by customer' and references OpenAI's similar situation.
The partial reinstatement of Mythos 5 reflects a compromise between rapid AI deployment and national security controls, with the Trump administration maintaining tight oversight. While the move is seen as progress for Anthropic and its clients, critics like Sam Altman caution against government picking winners in AI access. The fate of Fable 5 and broader policy implications remain open, as stakeholders watch for lasting frameworks.
Number of organizations gaining access to Mythos 5
| Outlet | Claim |
|---|---|
| Engadget | More than 100 institutions, including major corporations and government agencies. |
| DW English | More than 100 companies and institutions, including many Fortune 500 companies. |
| Wired | More than 100 US organizations, including large corporations and government agencies. |
The partial lift on Mythos 5 represents a significant but cautious step in US AI governance, balancing commercial interests with national security. The differing tones across outlets reflect broader political divides: Engadget and Wired report the facts with minor commentary, while DW injects skepticism about government control. The omission of detailed vetting criteria leaves room for concern about transparency, as noted by Altman. The unresolved status of Fable 5 suggests that the regulatory struggle is far from over.
Get tomorrow's top stories in your inbox