The US Commerce Department has removed from its website details of a deal under which Microsoft, Google and xAI agreed to submit new AI models to government scientists for security testing before their release, Reuters reported on Monday.
The original page, posted on May 5, said the three companies would hand over their border AI systems to the department’s testing team to review them for cyberattack vulnerabilities, risk of military misuse and national security flaws before public deployment.
On Monday afternoon Washington time, the link returned a “Sorry, we can’t find that page” error message; He was later redirected to the website of the AI Standards and Innovation Center, the government body that conducts the tests.
The Center, the successor body to the US AI Safety Institute (AISI), is located within the National Institute of Standards and Technology (NIST), part of the Department of Commerce.
The name change and reorientation followed an executive order that scaled back the previous administration’s AI safety architecture and reframed the institute’s mission around industry standards and coordination rather than safety assessment.
Reuters reported that neither the Commerce Department nor the Trump White House immediately responded to requests for comment on why the page was removed. There is no public statement from Microsoft, Google or xAI about the change.
The May 5 announcement was interpreted at the time as a notable commitment by the three companies to pre-deployment government review and as a sign of growing federal concern about the national security risks posed by powerful artificial intelligence systems.
The agreement followed The Trump administration’s elimination of Anthropic of a Pentagon AI contract for alleged security-related limitations; Anthropic was not named a participant in the Commerce Department’s testing program.
Deletion does not necessarily mean that the program has been cancelled. The AI Innovation and Standards Center continues to operate and the redirected web page suggests that the relationship between the agencies and the three companies remains at an operational level.
However, several federal officials have publicly questioned the wisdom of granting the government access to frontier AI models prior to launchbecause such access could become a target for cyberespionage by nation-states.
The story is more important as a sign. The Commerce Department’s willingness to remove a positive announcement about AI safety from its public website, without explanation, will be interpreted by both critics and supporters of US AI policy as evidence of internal disagreement over how the government should interact with frontier AI labs.
Industry observers had treated the original May 5 announcement as a stable element of the new administration’s policy stance on AI.
Microsoft, Google and xAI did not respond to Reuters requests for comment. Anthropic, OpenAI, Meta, and other large model vendors were not part of the original announcement and have not commented on the removal.
The AI Standards and Innovation Center website, where the redirect now points, contains general information about its program, but does not currently include the details of the pre-launch testing agreement that were on the removed Commerce page.
