AI-Ready Bio-Data Standards Act

This bill directs the Director of the National Institute of Standards and Technology (NIST) to establish definitions, standards, resources, and frameworks to ensure certain biological datasets generated from federally funded research are ready for use in artificial intelligence (AI) models. It also requires agency-specific data management policies, an advisory group, public input, testing, and reports to Congress.

Sec. 2, Definitions, standards, resources, and frameworks by NIST: Within two years, the NIST Director must establish definitions for "artificial intelligence-ready," "biomanufacturing," "biotechnology," and "qualified federally funded research." The definition of AI-ready must require datasets to be formatted for effective AI training and biotechnology research. The Director may determine a dataset is not AI-ready. Conditions for qualified federally funded research include funding amount, recipient capability, expertise, dataset size, and other factors. The Director must also set standards for making datasets AI-ready, and create data management resources and cybersecurity frameworks for federal agencies and researchers. These must not be overly burdensome and must be tested and evaluated. The Director must update them annually and consult with private sector entities, academia, and specified federal agency heads.

Sec. 2(b), Information gathering: Within one year, the Director must inventory existing biotechnology standards and biological datasets from federally funded research. Within one year after that, the inventory must be made public on a NIST website.

Sec. 2(c), Test and evaluation: Within one year and at least every two years thereafter, the Director must coordinate with the National Science Foundation to test and evaluate the definitions, standards, resources, and frameworks on sample datasets to assess clarity, ease of use, and whether compliance creates undue burden.

Sec. 2(d), Agency-specific data management policies: Within two years, the Director must establish or revise data management policies for each federal agency that funds qualified research, requiring that generated biological datasets be AI-ready. Policies must include sufficient funding for recipients, a compliance designee, a central public repository of policies, a public database for AI-ready datasets, a reporting mechanism, and a mechanism for agencies to request assistance.

Sec. 2(e), Public input and feedback; consultation: The Director must solicit public input and consult with specified federal agency heads to ensure agencies can comply with the definitions, standards, resources, frameworks, and agency-specific policies.

Sec. 2(f), Advisory group: Within 180 days, the Director must establish an advisory group of at least 12 members from federal agencies, academia, private sector, and academic publishers. The group provides recommendations on definitions, standards, resources, frameworks, and agency policies; recommends guidelines for academic journals; and solicits academic community input. Members serve two-year terms; the chair serves a one-year term. An interim report is due within one year, with subsequent reports as determined.

Sec. 2(g), Federal Acquisition Regulation revisions: The Federal Acquisition Regulatory Council must revise the Federal Acquisition Regulation as needed to implement the definitions, standards, resources, and frameworks.

Sec. 2(h), Annual report: Within one year, the Director must submit an interim report to Congress and the Comptroller General on progress. Within two years and annually thereafter, a report must cover establishment, implementation, and revision of definitions, standards, resources, frameworks, and agency policies. The first subsequent report must include a summary of testing and evaluation, findings, how concerns were addressed, burden assessment, and a cost-benefit analysis.

Sec. 2(i), Government Accountability Office report: Within five years, the Comptroller General must report to Congress on the impact of the definitions, standards, resources, and frameworks, including effectiveness, undue burden, and recommendations.

Sec. 2(j), Sunset: This section terminates 10 years after enactment.

Sec. 2(k), Definitions: "Biological data" means information derived from the structure, function, or process of a biological system. "Biological dataset" means a discrete collection of biological data. "Biological data repository" means centralized data storage for biological data. "Negative data" means data that disproves or fails to support a hypothesis but advances scientific knowledge.