US permits Anthropic to share Mythos model with trusted partners

The United States has granted Anthropic permission to share its Mythos large‑language model with a limited set of trusted partners, according to Reuters citing a Semafor report [Reuters]. The clearance allows Anthropic to provide the model to selected U.S. companies that meet the government’s “trusted partner” criteria, giving them direct access to the next‑generation LLM for product integration.

The decision establishes a regulatory framework that defines how advanced AI models can be distributed. It signals to the industry that the government is willing to approve controlled releases, offering clear guidance on compliance requirements. The trusted‑partner approach creates a structured pathway for responsible sharing, ensuring that sensitive technology remains under oversight while still enabling commercial use.

By opening Mythos to a handful of vetted firms, the policy could shift the competitive dynamics of the AI market. Companies with early access gain a head‑start in developing AI‑driven services, potentially accelerating investment and innovation across the sector. The move also provides a template for future model releases, showing how regulatory bodies can balance safety concerns with industry growth.

[Reuters] confirms that the permission applies to “some US companies,” underscoring the targeted nature of the rollout and the emphasis on responsible collaboration.