Google, Microsoft, xAI join US AI review program

Google, Microsoft, and xAI have agreed to let the US Commerce Department review some AI models before public release.

The move expands a federal testing program tied to Donald Trump’s AI policy.

They join OpenAI and Anthropic, whose models the department has reviewed since 2024 through its Center for AI Standards and Innovation, formerly the AI Safety Institute.

The office said it has completed more than 40 evaluations, including unreleased systems.

The agreements were announced on May 5 as the White House considers a broader review process for AI tools.

Trump’s July AI Action Plan directs the center to lead national security assessments and explore how evaluations could support enforcement under existing law.

🔗 Source: Bloomberg

🧠 Food for thought

Implications, context, and why it matters.

The Commerce Department’s AI review center now has a more pro-innovation mandate

The Center for AI Standards and Innovation was renamed from the U.S. AI Safety Institute in June 2025 under the Trump administration 1.
The rename marked a policy turn. The Commerce Department framed the center as a way to back American interests abroad and push back on “burdensome and unnecessary regulation” from foreign governments 1.
The office now reviews commercial AI systems with an eye on U.S. innovation and national security. “Innovators will no longer be limited by these standards,” said Howard Lutnick, commerce secretary 1.

Joining model reviews does not erase wider government-industry strains

Google and Microsoft have joined the program. Anthropic, an original participant, is in a legal fight with the Trump administration over restrictions tied to use of its technology 2.
The Department of Defense (DOD) labeled Anthropic a “supply chain risk” in early March 2026 after a dispute over limits Anthropic wanted on Claude, its AI model, for fully autonomous weapons and mass surveillance of Americans 3.
Taking part in model evaluations does not shield companies from later clashes over how the government uses AI systems or limits them 4.
A federal appeals court refused Anthropic’s bid to temporarily block the Pentagon’s blacklist. That leaves the company out of DOD contracts while the case moves ahead 3.