AI models: Trump to test Google, Microsoft and xAI before release

The U.S. administration has reached agreements with Google DeepMind, Microsoft and Elon Musk’s xAI to evaluate new artificial intelligence models before they are publicly released. The deals expand the government’s ability to conduct pre-deployment security and national-security assessments of frontier AI systems.

The Center for AI Standards and Innovation (CAISI) at the Department of Commerce will run early-access evaluations and targeted research on the companies’ frontier models. CAISI says it has already completed more than 40 such evaluations, and the new agreements bring Google, Microsoft and xAI into the same pre-release testing framework that previously included OpenAI and Anthropic.

Financial markets could reprice regulatory and deployment risks for major tech firms. Early-access reviews introduce the possibility of delays or additional safeguards before model rollouts, which may affect the timelines and revenue forecasts for cloud and AI product lines at Microsoft (MSFT) and Alphabet/Google (GOOGL). Investors are likely to react to agency statements and any concrete findings from CAISI assessments.

The move follows heightened concern around Anthropic’s recently unveiled Mythos model and its purported ability to surface software security vulnerabilities. That episode prompted federal officials to consider more formal review mechanisms, and reporting indicates the White House is weighing an executive order that could formalize pre-release vetting for high-risk AI systems.

Analysts say the near-term market impact will depend on the cadence and transparency of CAISI testing and any subsequent policy steps. Over the medium term, standardized government testing could make AI deployments more predictable for enterprise customers while raising compliance costs for developers. Key indicators for investors will be CAISI’s technical reports, company disclosures on model timelines, and any enacted executive actions.