H.R. 9341: AI-Ready Federal Data Guidelines Act

This bill would direct the National Institute of Standards and Technology (NIST) to create voluntary guidelines for federal agencies on how to prepare government datasets, including open government data, so they can be used to train artificial intelligence models.

What the guidelines would cover

The guidelines would be meant to help agencies make data more useful for AI by addressing topics such as:

• Data formatting and structure so AI systems can better interpret the data
• Data labeling and annotation, including automated and expert-assisted methods
• Data quality checks to judge whether a dataset is suitable for AI use
• Metadata and documentation so users understand what the data means and how to use it
• Ongoing maintenance of datasets so they stay useful over time
• Access to public data for AI development and use

The bill says the guidelines should be flexible enough to work across different sectors and scientific fields, and should generally align with existing federal standards and conformity assessment procedures where practical.

Pilot programs

The bill would allow NIST to run pilot programs to develop and test AI-ready data practices in specific sectors or scientific areas. These pilots would be limited to one year each.

If used, the pilot programs would:

• Develop additional guidance for specific sectors or research areas
• Measure whether the guidelines improve data usability, interoperability, and AI readiness
• Identify technical, operational, or resource challenges
• Develop a path for handing future maintenance and updates to a non-federal entity where appropriate

The pilots would prioritize areas with major national security or industrial competitiveness importance, such as biotechnology and biomanufacturing, and areas where federal agencies already control and maintain AI-ready datasets. NIST could run no more than two pilot programs at the same time, through federal research programs, National Laboratories, universities, or private-sector partnerships.

Reporting to Congress

NIST would have to brief House and Senate committees within one year after the guidelines are published, and then once a year for five years afterward, on how the section is being implemented.

Funding restriction

The bill says NIST could not take money away from other programs or activities to pay for this work.

Definitions and related change

The bill defines terms like “artificial intelligence,” “artificial intelligence model,” “artificial intelligence system,” “open Government data asset,” and “conformity assessment procedure.” It also makes a conforming change by repealing a related subsection in current law.

Relevant Companies

None found.