The U.S. Commerce Department has removed from its website the details of a May 5 agreement with Google, Microsoft, and xAI to test their artificial intelligence models for security risks. The webpage that announced the partnership, which would allow government scientists to vet new AI models before their public release, was taken down on May 11.
The now-deleted announcement detailed a plan for the companies to provide early access to their models to the Center for AI Standards and Innovation (CAISI), part of the National Institute of Standards and Technology. The goal of these collaborations is to identify national security risks, such as potential for cyberattacks or misuse, before the technology becomes widely available. It was not immediately clear why the details of the agreement were removed from the website.