Key Highlights
- Anthropic’s Fable 5 AI system may resume operations following a two-week government-imposed suspension
- Final authorization from the Pentagon and NSA remains pending before full restoration
- Limited access to Mythos 5 was reinstated Friday by the Commerce Department for select verified users
- Commerce Secretary Howard Lutnick and Treasury Secretary Scott Bessent facilitated resolution efforts
- Anthropic and OpenAI are advocating for establishing standardized government evaluation procedures for cutting-edge AI systems
According to reporting from Axios, Anthropic’s Fable 5 AI system could resume full operations within days. The Trump administration appears ready to remove the restrictions that have kept the technology unavailable since its shutdown on June 12.
The suspension occurred after U.S. authorities implemented export control measures based on national security considerations. This action left numerous developers and commercial entities without access to a resource that had become integral to their operations.
Sources with knowledge of the ongoing negotiations told Axios that the restrictions might be removed as soon as the upcoming week. Discussions between Anthropic representatives and government officials are scheduled to continue throughout the weekend.
However, universal approval has not yet been achieved. Both the Pentagon and the National Security Agency must provide their authorization before the system can be reactivated. Several other federal agencies have already determined that the model presents no unacceptable risk to public deployment.
Commerce Secretary Howard Lutnick and Treasury Secretary Scott Bessent were instrumental in advancing the resolution. In correspondence to Anthropic, Lutnick stated that the organization “has worked with the US government to address risks” related to both AI systems.
Partial Restoration of Mythos 5 Completed
The Commerce Department granted Anthropic authorization on Friday to reinstate Mythos 5 access for a select group of vetted users. Mythos 5 represents the more sophisticated of the two systems and has never been released to general audiences.
Both the Fable 5 and Mythos 5 systems share the same foundational AI architecture. The primary distinction lies in their deployment strategy: Fable 5 targets widespread public accessibility, whereas Mythos 5 incorporates enhanced security protocols to minimize risks such as cyber exploitation or bioweapon development.
The Significance of Fable 5 for Development Teams
Prior to its June 12 suspension, Fable 5 had gained significant traction among software engineers due to its sophisticated coding and analytical capabilities. Financial technology firm Stripe allegedly utilized the system to restructure a massive 50 million-line codebase in just 24 hours—a project that would have required manual engineering efforts exceeding two months.
Following the suspension, automated development processes were interrupted, prompting some organizations to migrate their operations to alternative AI platforms, including more affordable Chinese-developed models.
The shutdown emerged amid broader tensions between Anthropic and the Trump administration. Defense Secretary Pete Hegseth had previously characterized Anthropic as a “Supply-Chain Risk to National Security.” The anticipated reinstatement of Fable 5 signals an improvement in this strained dynamic.
An administration official informed Axios that Anthropic “has worked positively with the government.”
Demands for Standardized Evaluation Framework
Both Anthropic and OpenAI are urging the Trump administration to establish a formalized evaluation framework for advanced AI technologies prior to their public release. This initiative follows President Trump’s June 2 executive directive introducing optional government assessment for high-capability AI systems.
OpenAI secured authorization Friday for a restricted preview of GPT-5.6. In a company statement, OpenAI indicated it does not believe government access protocols “should become the long-term default.”
Anthropic has similarly advocated for an evaluation process that is “transparent, fair, clear, and grounded in technical facts.”
