Commerce Department to Pre‑Test AI Models from Google, Microsoft and xAI

Context The Center for AI Standards and Innovation (CAISI), housed in the Commerce Department, has secured voluntary agreements with three major AI developers. The move expands earlier pacts with OpenAI and Anthropic and reflects a shift toward pre‑emptive safety checks amid growing public and military use of generative AI.

Key Facts - CAISI will evaluate Google’s Gemini chatbot, Microsoft’s CoPilot assistant and xAI’s Grok chatbot before they become publicly available. - Director Chris Fall said the expanded collaborations “help us scale our work in the public interest at a critical moment.” - The centre has already completed 40 evaluations of AI tools, including tests of unreleased state‑of‑the‑art models. - Microsoft’s corporate blog notes that national‑security and large‑scale safety testing requires government partnership. Google’s DeepMind declined comment, and xAI did not respond.

What It Means Pre‑release testing gives the government a chance to assess capabilities, security flaws and potential misuse before models reach consumers or defense contracts. By covering “testing, collaborative research and best practice development,” the program aims to set baseline safety standards across the commercial AI sector. The inclusion of Google, Microsoft and xAI signals that the administration is moving beyond the hands‑off stance of the previous administration, which prioritized deregulation.

The approach could influence future regulatory frameworks and may prompt other AI firms to submit models for review. Watch for the first set of test results and any guidance that emerges on model deployment restrictions.