The bill would require federal agencies to follow the National Institute of Standards and Technology’s (NIST) AI risk guidance when buying or deploying AI systems.
Representatives Ted Lieu (D-Calif.), Zach Nunn (R-Iowa), and Don Beyer (D-Va.) introduced a bipartisan bill, H.R. 8819, that would require federal agencies to use NIST’s Artificial Intelligence Risk Management Framework when developing, buying, or using AI systems. The bill would also require agencies to work with NIST to create additional AI guidance and training programs for federal employees, excluding national security agencies.
What is the NIST Artificial Intelligence Risk Management Framework?
It’s a set of U.S. government guidelines that helps organizations safely and responsibly develop, buy, and use AI systems. It provides companies and government agencies with a practical process for identifying and reducing AI-related risks. The framework has four main areas:
- Govern — Create internal policies, assign responsibility, and establish oversight for AI use.
- Map — Identify where AI is used, what data it relies on, who it affects, and what risks it creates.
- Measure — Test the system for problems such as inaccurate results, security flaws, bias, and reliability issues.
- Manage — Put safeguards in place, monitor performance, respond to failures, and update systems over time.
The bill follows several recent government actions related to AI procurement and oversight by federal agencies. In April 2025, the White House Office of Management and Budget (OMB) issued guidance directing agencies to establish minimum risk management practices for “high-impact AI” systems such as healthcare, housing, employment, and access to government services.
Unlike earlier Biden administration guidance, the current OMB memoranda do not uniformly require agencies to follow the NIST guidelines. H.R. 8819 would reverse that approach by requiring agencies to use the NIST framework government-wide.
Recently, NIST expanded its role in evaluating AI systems used by federal agencies. In March, the General Services Administration announced a partnership with NIST’s Center for AI Standards and Innovation (CAISI) to test AI systems for security and performance risks before deployment.
The bill does not create new penalties for agencies that fail to comply. It also does not explain how compliance would be audited or enforced.
The House Science, Space and Technology Committee is reviewing for further consideration.
