US to Put Google, Microsoft, and xAI AI Models Through Safety Vetting
Under a voluntary expansion of Biden-era accords, the US Department of Commerce will test AI models from Google (Gemini via DeepMind), Microsoft (Copilot), and xAI (Grok) through the Center for AI Standards and Innovation to evaluate capabilities and security, with CAISI noting it has run about 40 prior evaluations and emphasizing collaborative research and best-practice development for commercial AI systems.