# Google, Microsoft, and xAI agreed to give the U.S. government early access to frontier models for national security testing.

*genai · news · 2026-05-05 · Decrypt*

## Key points

- Google, Microsoft, and xAI will give U.S. officials pre-release access to their frontier AI models.
- Anthropic's Claude Mythos model has demonstrated advanced capabilities in discovering cybersecurity vulnerabilities.
- The Trump administration is considering an executive order to mandate pre-release review of advanced AI models.
- Anthropic is currently restricting Claude Mythos access to select organizations amid national security concerns.
- The Trump administration previously clashed with Anthropic over denied requests for unrestricted model access.

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

[Read the full story on Decrypt](https://decrypt.co/366810/us-government-vet-pre-release-ai-models-google-xai-microsoft)

---

Canonical: https://newsio.io/n/3ea053a9-cbc2-4cf1-bb9d-0539dd73f5df/google-microsoft-and-xai-agreed-to-give-the-u-s-government-early-access-to-front
Summarized by Newsio from Decrypt. https://newsio.io/how-it-works