Google, Microsoft, and xAI Agree to US Government Testing of AI Models for Security Risks

The US Department of Commerce announced on Tuesday that Google, Microsoft and xAI have agreed to submit their new AI models for testing before public release, expanding voluntary pacts to evaluate capabilities and security risks. The agreements, reached through the Center for AI Standards and Innovation (CAISI), build on prior arrangements with OpenAI and Anthropic from the Biden administration.

AI models from Google, Microsoft, xAI, OpenAI and Anthropic will undergo evaluations for national security, cybersecurity and biosecurity risks, according to announcements from the department and multiple sources.

CAISI has conducted more than 40 evaluations of AI tools, including state-of-the-art models that remain unreleased, the center stated. Evaluations will cover testing, collaborative research and best practice development for commercial AI systems. The pacts allow the US government pre-release access to the most advanced frontier AI models from Alphabet, Microsoft and xAI, led by CAISI.

Previous agreements with Anthropic and OpenAI have been renegotiated to align with Commerce Secretary Howard Lutnick and President Trump's new directives on security reviews. Tom Lue, vice president of global AI affairs at Google DeepMind, confirmed the partnership in a social media post on Tuesday.

The agreement was announced on 5 May 2026, amid a separate Pentagon deal with seven tech companies—Google, Microsoft, Amazon Web Services, Nvidia, OpenAI, Reflection and SpaceX—to use AI in classified systems.

Anthropic developed a model called Mythos, and is involved in a lawsuit with the US Department of Defense over its refusal to drop safety guardrails for government use of its models. Senior members of Trump's staff met last month with Anthropic CEO Dario Amodei. US President Donald Trump signed executive orders last year forming the basis of his administration's AI Action Plan.

The Trump administration pledged in July to partner with technology companies to vet AI models for national security risks. Agreements with OpenAI and Anthropic were made in 2024 under former President Joe Biden’s administration, when CAISI was known as the US Artificial Intelligence Safety Institute.

The AI Safety Institute was established in 2023 under the Biden administration and renamed CAISI under the Trump administration.

CAISI evaluated and tested certain state-of-the-art models that remain unreleased, completing over 40 such evaluations on Tuesday. A Pew Research Center poll last year found that 50% of Republicans and 51% of Democrats were more concerned than excited about increased use of AI in daily life. 3 percent.

Microsoft holds a Seeking Alpha Quant Rating of Strong Buy, and Google holds a Seeking Alpha Quant Rating of Hold.