Last progress May 7, 2025 (7 months ago)
Introduced on May 7, 2025 by Ben Ray Luján
Read twice and referred to the Committee on Commerce, Science, and Transportation.
This bill sets up a national effort to test and grade how well AI systems work, especially those used by federal agencies. The National Institute of Standards and Technology would run a pilot program with the Department of Energy to build “testbeds” — controlled places to safely try out AI and measure things like reliability, privacy, security, and bias. A small working group from government, universities, and industry would guide what to measure and how, and the plan would be posted publicly so developers and researchers can use it too . The goal is AI that is safer, more fair, and more trustworthy in services that affect people’s daily lives, like benefits processing, health, or transportation .
Key timelines and impacts:
| Who is affected | What changes | When |
|---|---|---|
| Federal agencies using AI | Common tests and standards for AI | MoU within 180 days; review every 2 years |
| AI developers, researchers, industry | Public strategy and testbeds to guide safer, fairer AI | Strategy within 1 year of group forming |
| General public | More trustworthy AI in government services | Testbeds within 2 years; report 180 days after first demo |
Overall, this creates clear rules of the road for testing AI used by the government, aiming to reduce errors and bias, protect privacy, and improve performance before these tools are used in ways that affect people’s lives .