The Trump administration is close to finalizing an AI voluntary review framework

July 28th, according to The Information, three sources revealed that the Trump administration is about to finalize a voluntary framework requiring AI companies to submit their cutting-edge models to the government before releasing them to the public. The sources said that about two weeks ago, the White House National Network Office distributed a draft of the framework to OpenAI, Anthropic, and Google (GOOG.O), and the three companies each submitted their proposed modifications. An executive order on AI signed by Trump in early June set August 1st as the deadline for creating this framework, which aims to establish a more standardized review process for powerful new AIs capable of quickly identifying network vulnerabilities.

According to two sources, as the deadline approaches, some AI companies have expressed concerns during meetings with White House agencies about how the framework will define cutting-edge models and whether it will differentiate between closed-source and open-source models. An unresolved issue is how voluntary the framework will actually be in practice, especially after the White House decided to impose export controls on Anthropic. The Trump administration insists that compliance with this framework is not mandatory.