US Government to Vet AI Models from Google, Microsoft, xAI

Google, Microsoft, and xAI will give the U.S. government early access to their AI models for national security testing. This comes as President Trump considers an executive order for AI oversight. The Commerce Department's Center for AI Standards announced this collaboration, highlighting the need for rigorous evaluation of AI technologies. Chris Fall, the center's director, emphasizes the importance of understanding AI’s implications for national security. The discussions were partly spurred by concerns over Anthropic's new Claude Mythos model, which can identify cybersecurity vulnerabilities. Instead of a broad public release, Anthropic limits access to select organizations, allowing Mozilla to patch 271 vulnerabilities in its Firefox browser using Mythos. In a twist, the Trump administration previously clashed with Anthropic over access to its models, leading to a designation of supply chain risk for national security. Now, the administration is reconsidering its partnership with Anthropic. The bottom line is that how the government handles AI oversight could shape the future of technology and security for everyone.