ChainThink reports that on August 4, the Trump administration completed a review framework for the release of advanced AI models and invited staff from OpenAI, Google, and Anthropic to the White House for discussions.
Under this framework, AI companies may voluntarily submit their models to the government for testing before making them available to partners and the public. The framework stems from an executive order signed by Trump on June 2, which required multiple agencies to develop rules within 60 days identifying which models require heightened scrutiny;
Previously, Anthropic did not publicly release its advanced model, Mythos, due to concerns about cybersecurity risks. The government seeks to collaborate with companies to identify cyberattack and national security risks before model releases, but does not currently intend to establish a mandatory approval system.
The White House stated that the framework has been completed, but has not yet indicated whether it has officially taken effect. Previously, the U.S. government restricted the export of Anthropic’s Fable model and required OpenAI to release GPT-5.6 in phases.
The new framework will incorporate relevant temporary interventions into fixed procedures.
