Key facts
- Anthropic's Mythos 5 AI model approved for limited rollout by Trump administration
- Model had been blocked previously over national security concerns
- US negotiated security safeguards as part of the deal
- Full wide release was not approved — access remains restricted
The Trump administration has approved a limited rollout of Anthropic's Mythos 5 AI model, easing export restrictions that had been imposed over national security concerns — but stopping well short of a full public release. The US government and Anthropic moved toward a formal deal that permits restricted access to the model, with security safeguards built into the arrangement, according to multiple reports.
The model had previously been blocked from wider deployment after the US government raised national security concerns, making Mythos 5 one of the most closely scrutinised AI releases in recent memory. The limited approval means access will be controlled rather than open, with the Trump administration keeping a tight grip on how and to whom the technology is made available.
The agreement reflects Washington's increasingly hands-on posture toward frontier AI companies. Rather than letting the market decide the pace of deployment, the US government is actively brokering terms with AI labs — a dynamic that is reshaping how even private AI products reach the world. Anthropic, which positions itself as a safety-focused AI company, appears to have worked within the government's framework to secure at least partial clearance.
For the broader AI industry, the episode underlines a new reality: the most powerful models may no longer be purely commercial launches. Regulatory approval — at least informally — is becoming part of the rollout process for top-tier AI systems, with national security considerations now sitting alongside capability benchmarks as a gating criterion.