US Government Gains Early AI Access

Microsoft, Alphabet Inc., and Xai have agreed to provide the United States government with early access to their new artificial intelligence models for national security testing. This agreement, announced by the United States — United States Department of Commerce's Center for AI Standards and Innovation (CAISI), aims to evaluate AI models before deployment and research their capabilities and security risks. The move fulfills a pledge made by the Donald Trump administration in July 2025. Concerns about the national security risks of powerful AI systems, particularly the hacking capabilities of Anthropic's Mythos model, have driven this initiative. Microsoft will collaborate with US government scientists to test AI systems for unexpected behaviors and has a similar agreement with the United Kingdom's AI Security Institute. This builds on previous agreements with OpenAI and Anthropic under the Joe Biden administration. United States — The Pentagon also recently reached agreements with seven AI companies, including Alphabet Inc., Microsoft, Amazon — Amazon Web Services, Nvidia, OpenAI, Reflection, and SpaceX, to deploy AI on classified networks, notably excluding Anthropic due to an ongoing dispute.