The US government has reversed its earlier decision to block Anthropic's powerful Claude Mythos 5 AI model. After a two-week suspension triggered by an export control directive, Commerce Secretary Howard Lutnick notified Anthropic that 'appropriate safeguards are in place' to allow the model's redeployment to a select group of US organizations. These include over 100 entities such as Fortune 500 companies, federal agencies, and cybersecurity firms that operate critical infrastructure. Anthropic had abruptly suspended access to Mythos 5 and its consumer counterpart Fable 5 on June 12 following government orders to restrict access by foreign nationals. The company engaged in daily talks with the administration to address security concerns, which included fears that the model could be jailbroken or used by China-linked groups. While Mythos 5 is being restored to trusted partners, Fable 5 remains blocked pending further discussions. The decision comes amid broader tensions between the Trump administration and AI companies. OpenAI's GPT-5.6 is also subject to similar restrictions, and OpenAI CEO Sam Altman publicly criticized the idea of the government picking which customers can access frontier models. The incident has sparked debate about national security versus industry autonomy in AI development.
Revocation of block as easing of tension between Trump administration and Anthropic
Il Fatto reports the unblocking as a sign of reduced confrontation, highlighting the letter from Lutnick and Anthropic’s commitment to cooperation. It also notes the simultaneous release of OpenAI's GPT-5.6 to approved partners.
Dutch perspective: AI tool allowed for government agencies after safety reassurances
NOS explains that Mythos can now be shared with certain government agencies to check for security vulnerabilities. It notes that the block was initially due to fears of misuse against the government, and that talks are ongoing to restore full access to Fable.
Wired reports the partial reinstatement as a step forward but notes the government stopped short of a broader rollout and said nothing about Fable 5. It highlights that foreign national employees of approved organizations and Anthropic itself may now access the model.
Partial release amid security fears and government overreach accusations
DW covers the announcement as a partial restoration, emphasizing the 'thorny relationship' between Anthropic and the Trump administration. It includes criticism of government overreach and quotes Sam Altman's concerns about the government picking customers.
Engadget focuses on the permission to redeploy Mythos 5 to organizations defending critical infrastructure. It details the background of the block, including warnings from Amazon and concerns about a China-linked group accessing the model.
The partial unblocking of Anthropic's Mythos 5 represents a negotiated compromise between US national security concerns and the AI industry's push for deployment. While the administration has allowed access to vetted domestic partners, the continued restrictions on Fable 5 and the wider rollout underscore the government's cautious approach. The episode highlights the evolving tension between innovation and security in frontier AI, with unresolved questions about how future model releases will be governed.
Status of Fable 5
| Outlet | Claim |
|---|---|
| Il Fatto Quotidiano | Fable 5 should be unblocked shortly. |
| DW English | Fable 5 remains blocked; talks continue. |
| Wired | Fable 5 still blocked; talks ongoing over the weekend. |
The exact number of organizations granted access to Mythos 5
| Outlet | Claim |
|---|---|
| Il Fatto Quotidiano | Over 100 institutions, including large enterprises and federal agencies. |
| DW English | More than 100 companies and institutions, including many Fortune 500 companies. |
| Engadget | More than 100 institutions in the US, including major corporations and government agencies. |
| Wired | More than 100 US organizations, including large corporations and government agencies. |
The coverage reveals a common narrative: the US government is cautiously allowing limited access to Anthropic’s powerful cybersecurity model while maintaining strict controls. The partial unblocking is a win for Anthropic’s diplomacy but leaves the fate of the wider release uncertain. The framing differences reflect each outlet's focus—some highlight government-industry tensions, others emphasize security or technical details. The omission of specific safeguards and vetting criteria leaves room for future reporting on the policy implications.
Recibe las mejores historias de mañana en tu correo