WASHINGTON — The Trump administration is reviewing early access to powerful artificial intelligence models under a recently signed executive order that establishes a voluntary framework for companies to share frontier AI systems with the government before broader release.
President Donald Trump signed the order on June 2, directing federal agencies to design a program allowing AI developers to provide access to certain advanced models for up to 30 days prior to release to trusted partners. The measure aims to address national security and cybersecurity concerns associated with advanced AI capabilities.
The executive order calls for a benchmarking process to identify “covered frontier models” based on their advanced cyber capabilities. It also enables collaboration between developers and the government to select trusted partners for early access, subject to confidentiality and intellectual property protections.
“This framework promotes innovation while strengthening security,” a White House official said, according to statements accompanying the order. Details of how the voluntary framework will be implemented remain subject to further agency guidance.
The order follows earlier discussions and agreements with companies including Google, Microsoft and xAI that allow government review of new models. A previous version of the proposal, which reportedly included a longer 90-day review period, was shelved amid industry input.
Industry groups have expressed support for balanced approaches that protect proprietary information while addressing risks. Some developers have already participated in voluntary testing arrangements with agencies such as the National Institute of Standards and Technology.
Critics have raised concerns about potential delays in innovation and the implications for competitive advantages. Supporters argue that early government insight into high-risk models is necessary to safeguard critical infrastructure.
The administration has emphasized a light-touch regulatory stance overall, promoting AI development while focusing on specific security threats. Agencies including the Commerce Department, Homeland Security and others are involved in the framework’s development.
Implementation deadlines include agency actions within 60 days of the order for certain components. The full scope of participation by AI companies is expected to evolve based on engagement with the voluntary program.
As of June 11, 2026, the Trump administration continues work on the review framework. Further guidance and potential company announcements regarding participation are anticipated in the coming weeks.
