# Google, Microsoft and xAI will share unreleased versions of their AI models with the government.

*genai · news · 2026-05-05 · ABC17News.com*

## Key points

- Google, Microsoft, and xAI will share unreleased AI models with the US government for evaluation.
- The Center for AI Standards and Innovation can now assess AI models' national security impact before public release.
- Anthropic’s Mythos model is restricted to select organizations due to its advanced cybersecurity capabilities.
- The White House is considering a formal review process for new AI models, departing from previous policy.
- CAISI has already completed over 40 AI model evaluations using these expanded partnerships.

**Companies:** Google, Microsoft, xAI, Anthropic, OpenAI
**Countries:** United States

[Read the full story on ABC17News.com](https://abc17news.com/money/cnn-business-consumer/2026/05/05/microsoft-google-and-xai-will-let-the-government-test-their-ai-models-before-launch/)

---

Canonical: https://newsio.io/n/a388cdc9-ea2c-4725-b135-04e4e72b3df4/google-microsoft-and-xai-will-share-unreleased-versions-of-their-ai-models-with
Summarized by Newsio from ABC17News.com. https://newsio.io/how-it-works