Bottom line: The White House met its August 1 deadline to finalize a voluntary framework for reviewing advanced AI models, though the content and publication status of the document have not yet been made public.
The US government says it has finalized a framework for reviewing advanced AI models. Details or a possible publication have not yet been announced.
According to a US government official, a framework is now in place that is meant to govern how advanced AI models will be reviewed going forward. The draft is set to be presented to leading AI companies this Tuesday at a meeting with the Office of the National Cyber Director (ONCD). This is according to four people familiar with the planning who spoke on condition of anonymity, as the discussions are internal. Among the expected participants are Meta, Anthropic and Google, according to three of the sources. Google, OpenAI and Anthropic had already jointly reviewed an earlier draft in late July and submitted proposed changes, as Politico previously reported.
The basis is an Executive Order from June 2, which set a 60-day deadline for completing the voluntary framework — that is, August 1. A government official confirmed to Politico, on condition of anonymity, that this deadline was met: the voluntary framework was completed on schedule, and discussions with industry on next steps are already underway. The document has not yet been publicly released, and the Executive Order itself does not make any binding statement on this point.
For compliance officers at companies outside the narrow circle of providers already involved, this creates practical uncertainty: since this is a voluntary regime, the question arises as to how companies that were not part of the White House briefing could even join the framework, should it not be published more broadly. The framework is meant to define how the federal government works with leading US AI developers before and after the deployment of particularly capable models — against the backdrop of the Trump administration’s aim to address safety and national security risks while otherwise pursuing a largely light-touch regulatory approach to the technology.
Meta, Anthropic, Google and OpenAI did not immediately respond to requests for comment on the matter. For compliance practice at companies with US ties, it remains unclear for now whether and in what form concrete review criteria, reporting obligations or deadlines will emerge from the framework, as long as no public text is available.
Source: www.politico.com · Published August 3, 2026
Lumi AI News — AI-assisted curation in accordance with Art. 50 EU AI Act. Paraphrasing and classification by Lumi News Pipeline v1.8.3.