US government agency to safety test frontier AI models before release

The Center for AI Standards and Innovation (CAISI), a NIST division under the US Department of Commerce, has signed pre-deployment testing agreements with Google DeepMind, Microsoft, and xAI — joining earlier agreements with Anthropic and OpenAI. The agreements allow CAISI to evaluate frontier AI models before public release. This follows reports that the White House is preparing an executive order to create a formal AI vetting system, partly driven by concerns over Anthropic's Mythos model and its ability to discover network vulnerabilities. Analysts see this as a significant policy shift from a previously hands-off regulatory stance, with implications for how AI vendors balance speed-to-market with growing government oversight requirements.