After a two-week negotiation with the Trump administration, Anthropic’s Mythos 5 AI model has been granted restricted redeployment to a select group of cybersecurity defenders and infrastructure providers, marking a crucial step amid ongoing regulatory controls.
- Mythos 5 access restored for approved cybersecurity and infrastructure groups.
- Public rollout of Fable 5 model remains delayed with no set timeline.
- Anthropic and US government developing AI release protocols amid export controls.
What happened
Anthropic’s Mythos 5 model, known for its cybersecurity capabilities, has been allowed to resume operations for a limited set of organizations after a prolonged negotiation with the Trump administration. The US Commerce Secretary notified Anthropic of revised licensing terms following the company’s risk mitigation efforts in collaboration with government officials.
The new arrangement permits certain trusted partners including cybersecurity defenders and infrastructure providers to access Mythos 5, despite ongoing export controls that restrict foreign nationals from using the model broadly. This selective access mirrors the approach recently adopted for OpenAI’s GPT-5.6, creating a closed preview environment for both leading AI labs.
Why it matters
This partial reopening represents a delicate balance between advancing AI innovation in cybersecurity and addressing national security concerns through strict export controls. Anthropic’s limited access enables vital US government agencies and trusted enterprise partners to leverage Mythos 5’s capabilities, which had previously been restricted, including by concerns over unauthorized foreign access.
The move responds to pressure from industry stakeholders wary of US AI leadership losing ground to global competitors, especially China, amid restrictive regulatory measures. It also highlights the challenges of governing AI technologies that have rapid development cycles and broad international interest, requiring nuanced policies that safeguard security without stifling progress.
What to watch next
Future developments will hinge on ongoing negotiations between Anthropic and the US government to expand Mythos 5’s access beyond the current limited group and to release the public-facing Fable 5 model. Both AI developers emphasize the importance of creating repeatable, transparent processes for future AI model licensing and deployment.
Observers should monitor how the Trump administration refines export control policies and implements the anticipated cybersecurity Executive Order framework to balance innovation with national security. The broader AI ecosystem, including enterprises and global partners, awaits clearer guidance on timing and conditions for general availability of cutting-edge AI tools.