Home » White House Yet to Release Finalized AI Testing Framework

White House Yet to Release Finalized AI Testing Framework

By

–

White House Building

WASHINGTON, September 14 — The White House still has not released the voluntary framework for testing frontier artificial intelligence models that it finalized in August, and there is no sign of when it might. That silence is the story now. The framework was completed by the start of August, following an executive order President Trump signed in June.

That order asked AI companies — Anthropic and OpenAI among them — to voluntarily give the federal government access to their most advanced models up to 30 days before release, with the aim of strengthening the models’ security. Trump had pulled the plug on an initial signing of the executive order, saying he did not want the federal government to slow down innovation.

The revised order went through. The framework built on top of it has stayed out of public view.

A review the public cannot see

Because the framework guiding how new models are evaluated remains confidential, it is not clear what standards the federal government is asking companies to meet — or whether firms are disclosing new breakthroughs at all. Neither the administration nor the companies are required to release the results of the reviews, or even to say whether they took part.

That opacity has drawn criticism from the Center for Democracy & Technology, which sent a letter to the White House urging the Trump administration to release its framework for review of frontier models. Tim Harper, who focuses on election security issues at the Center for Democracy & Technology, said, “The public has a right to know what the government’s interpretation of the law is on these issues.” The reporting on the unreleased framework comes from CBS News, dated September 10.

What’s changing underneath

The shift here is not a single policy reversal. It is a review process that runs on voluntary participation, confidential standards and no disclosure requirement — a combination that leaves outsiders unable to judge how frontier models are being assessed before they reach the public. Watch this space.

The open question is whether the administration publishes the framework at all, and whether the companies that opted in ever say so.

Sources

More stories by

Most read

AI Tools

Artificial Intelligence

Business

World News