The US Commerce Department removes details regarding security tests conducted by Microsoft, Google, and xAI.

      The US Commerce Department has removed from its website the specifics of an agreement where Microsoft, Google, and xAI had agreed to present new AI models for security testing to government scientists prior to their release, according to a report by Reuters on Monday.

      The initial page, which was published on May 5, stated that the three companies would submit their advanced AI systems to the department's testing team for evaluation of potential vulnerabilities to cyberattacks, risks of military misuse, and flaws related to national security before public launch.

      By Monday afternoon Washington time, the link displayed a "Sorry, we cannot find that page" error message and was later redirected to the website of the Center for AI Standards and Innovation, the government agency responsible for conducting the tests.

      The Center, which succeeded the US AI Safety Institute (AISI), operates under the National Institute of Standards and Technology (NIST), part of the Department of Commerce.

      This rebranding and shift in focus followed an executive order that scaled back the previous administration's AI safety framework and redefined the institute's mission towards standards and industry coordination rather than safety assessments.

      Reuters indicated that neither the Commerce Department nor the Trump White House immediately responded to inquiries regarding the page's removal. Microsoft, Google, or xAI have not publicly commented on the matter.

      The announcement on May 5 was interpreted at the time as a significant commitment by the three companies to undergo governmental review prior to deployment, reflecting heightened federal concern over national security risks associated with advanced AI systems.

      This agreement came after the Trump administration removed Anthropic from a Pentagon AI contract due to concerns related to safety; Anthropic was not mentioned as a participant in the Commerce Department's testing program.

      The deletion of the announcement does not necessarily imply that the program has been canceled. The Center for AI Standards and Innovation continues to function, and the redirected webpage indicates that the relationship between the agencies and the three companies still exists at an operational level.

      However, several federal officials have publicly questioned the prudence of granting the government access to advanced AI models before their release, as such access might become a target for state-sponsored cyber espionage.

      This situation is significant as an indication. The Commerce Department's decision to remove a positive announcement about AI safety from its public website without explanation will likely be interpreted by both critics and supporters of US AI policy as evidence of internal disagreements regarding the government's approach to engagement with leading AI laboratories.

      Industry analysts had viewed the original May 5 announcement as a stable component of the new administration's AI policy stance.

      Microsoft, Google, and xAI did not reply to Reuters' requests for comments. Anthropic, OpenAI, Meta, and other major model providers were not included in the initial announcement and have not addressed the deletion.

      The website of the Center for AI Standards and Innovation, where the redirect now leads, contains general information about its program but does not currently incorporate the details of the pre-release testing arrangement that were on the now-deleted Commerce page.