The US government has approved Anthropic to redeploy Claude Mythos 5, allowing American organisations to use the model again for protecting critical infrastructure. This decision mirrors a separate permission granted to OpenAI to offer its GPT-5.6 Sol model to select customers. Under the new rules, non-US nationals working for Anthropic or employed by approved entities may also access Mythos 5. The company is currently negotiating with officials to restore broad availability for Fable 5, though no timeline has been set. OpenAI anticipates a few weeks for wider access to its model, following a suspension that began on 12 June after government orders blocked both Mythos 5 and Fable 5.
This regulatory shift matters because it signals a potential easing of restrictions on advanced AI tools within the US security sector. The approval suggests authorities are willing to allow specific, high-risk models back into environments that manage national infrastructure, provided strict access controls remain in place. It also highlights the ongoing tension between maintaining safety protocols and restoring functional capabilities for essential services.
- Access is currently limited to US-based organisations and cleared non-citizens
- OpenAI expects a similar rollout for GPT-5.6 Sol within a few weeks
- Both Mythos 5 and Fable 5 were previously blocked by government order
