The US government has authorized Anthropic to redeploy its powerful Claude Mythos 5 AI model to a select group of American organizations, including major corporations and government agencies, two weeks after blocking the model over national security concerns. The decision, conveyed in a letter from Commerce Secretary Howard Lutnick, cites significant progress in safety discussions with Anthropic. The company had suspended all access to Mythos 5 and its consumer version Fable 5 on June 12 after receiving a government directive to restrict access by foreign nationals. More than 100 institutions will regain access to Mythos 5, but the broader rollout and release of Fable 5 remain under review. Anthropic is continuing talks with the White House to expand access and inform future AI governance frameworks.
Inside story with exclusive letter, highlighting foreign national restrictions and lingering uncertainty
Wired obtains Lutnick's letter and reports that foreign national employees can now access Mythos, while the broader rollout and Fable 5 remain unresolved. It adds context about a South Korean telecom and China-linked group concerns.
DW highlights the Trump administration's concerns about China and Russia misusing the AI, and quotes Sam Altman criticizing government picking customers. It notes the thorny relationship between Anthropic and the administration.
NOS reports that certain government agencies can now use Mythos for security checks. It emphasizes the fear that the AI could be used against the government and notes ongoing talks for Fable's full release.
Engadget details the timeline of the block and restoration, including the role of Amazon and NSA warnings. It reports that Lutnick cited 'significant progress' in daily talks and that Anthropic is working to expand access.
Il Fatto reports the revocation of the block as an easing of tensions between Trump administration and Anthropic. It quotes Lutnick's letter and notes OpenAI's parallel limited release of GPT-5.6.
The partial lifting of the Mythos 5 block represents a cautious step toward reconciling AI innovation with national security concerns. While the government maintains control over distribution, Anthropic has demonstrated willingness to comply with safety protocols. The incident underscores the ongoing tension between private AI development and government oversight, with unresolved questions about the fate of consumer models like Fable 5 and the criteria for selecting trusted partners.
Whether the restored access includes only government agencies or also private corporations
| Outlet | Claim |
|---|---|
| NOS | Bepaalde overheidsinstanties krijgen nu toegang tot Mythos (certain government agencies get access). |
| Wired | More than 100 US organizations, including large corporations and government agencies. |
| Engadget | More than 100 institutions in the US, including major corporations and government agencies. |
Whether foreign national employees are now allowed to access Mythos
| Outlet | Claim |
|---|---|
| Wired | Approved organizations and Anthropic may allow foreign national employees to access Mythos. |
| DW English | No mention of foreign nationals; earlier block was over foreign access concerns. |
The partial clearance of Mythos 5 is a limited, conditional victory for Anthropic. The government has taken a targeted approach, allowing access only to handpicked entities deemed critical to national infrastructure. This reflects an ongoing struggle between AI companies' desire for broad deployment and the government's risk-averse posture, especially regarding foreign access. The absence of a clear path for Fable 5 and the lack of transparency about selection criteria leave significant uncertainty. The differing framings—from empowerment of cybersecurity to fear of government picking winners—underscore the deep political and security dimensions of cutting-edge AI governance.
Get tomorrow's top stories in your inbox