The US Commerce Department removes from a public webpage details of an agreement involving Microsoft, Google and xAI to undergo security testing for new “frontier” AI models before release, Reuters reports and other outlets reference. The original page, published on 5 May, said the companies would provide newly developed models to government scientists for evaluation of potential safety and security risks. According to the reporting, the page has since been deleted, removing previously published information about the terms and participants in the arrangement. The outlets describe the deletion as affecting the availability of the agreement’s posted specifics, rather than reporting any change to the underlying commitments themselves. The companies and the US government have not uniformly provided additional public clarification in the cited accounts. The removal comes after increased public attention on how companies and governments coordinate to assess risks from advanced AI systems prior to public deployment.