The White House has finished work on a voluntary framework for reviewing advanced AI models, meeting the deadline set out in the June 2 executive order. Although officials confirmed that the framework is now complete, they have not shared what is in it, who has reviewed it, or when companies will actually start using it. The administration’s choice to keep the framework private stands out, since it is expected to influence how the U.S. government deals with developers behind some of the most advanced AI systems in the country.
Major AI companies, lawmakers, safety groups, and U.S. allies are all keeping a close eye on the framework. Many expected more openness from the administration once the document was done. Instead, officials said only that conversations with the industry have started, leaving basic questions about how the voluntary process will work unanswered.
White House Confirms the Framework Is Complete
A White House official confirmed that the voluntary framework, directed by the June 2 executive order, was completed by the deadline. The official said talks with the industry on next steps are underway, and that those talks involve a wide range of partners rather than only Anthropic, OpenAI, and Google. The administration has not said who has seen the document or when companies are expected to begin following the process.
Ahead of the deadline, Anthropic, OpenAI, and Google reviewed a draft and submitted feedback. Reports indicate a further draft will be reviewed by AI companies at a staff-level meeting with the Office of the National Cyber Director (ONCD). Anthropic, Google, Meta, and OpenAI are expected at that meeting. The deadline fell on August 1, 60 days after the executive order was signed. The administration insists the work finished on schedule, but has not said whether the final framework will ever be shared with the public.
What the Framework Covers on Access and Security
The framework is meant to lay out how AI developers can work with the federal government to decide whether their models qualify for the voluntary review process. Based on what is known so far, the framework covers confidentiality, cybersecurity, insider risks, intellectual property protection, usage limits, and nondisclosure commitments where the government gets access to advanced AI models before they are released publicly. It is also expected to set out how the government could obtain access to certain models up to 30 days before launch, and which “trusted partners” receive the same early access. The framework should also clarify which “trusted partners” are allowed that same early access to these systems.
The June 2 order also makes clear that the process for benchmarking advanced cybersecurity features in AI models is classified. Another classified detail is the threshold for deciding which AI models fall under these rules. Only relevant developers and researchers are expected to be told those details. Even with these classified elements, the order does not say that the framework itself has to be classified. So, many lawmakers and observers assumed that at least some framework details would become public after its completion. But a White House official was clear that just because something is not classified does not mean it will be shared publicly.
NEW AI FRAMEWORK ANNOUNCED
— One America News (@OANN) August 4, 2026
The White House has completed a new framework for reviewing advanced AI systems. Officials say it will require testing before release and establish standards for cybersecurity and data protections.
For more updates, watch OAN on OAN Live or click the… pic.twitter.com/M8hI5Ns56t
Why Industry Participation Remains Unclear
Withholding the framework raises questions about how companies outside these private talks can take part in what is described as a voluntary process. Reports indicate the framework has not been widely shared with the industry, despite weeks of direct work with the White House on its contents. That leaves developers who do not get direct White House briefings unsure how to join the process if the document never goes public.
The new framework is set to guide how the federal government collaborates with top AI developers before and after advanced models are released. The administration says it is addressing the safety and national security concerns raised by powerful AI while maintaining a mostly light-touch approach to regulating the industry. Companies are seeking clarity as they plan future releases, particularly on whether their systems fall under these new rules and what obligations follow. For now, though, most answers are still confidential as discussions between the administration and the industry continue.
