Sponsors (12)
The bill creates a NIST-backed, voluntary documentation standard to improve AI transparency, safety, and interoperability and lower integration costs, but its voluntary nature, funding dependence, potential IP/security exposures, and added burden on small developers may limit adoption and impose costs on startups and researchers.
Researchers, developers, regulators, and the public gain a standardized, modular AI model documentation template (to be drafted and published by NIST within 12 months) that improves access to model provenance and performance.
Small businesses and nonprofits can adopt consensus-based guidelines to demonstrate due diligence and interoperability, reducing integration costs when using or combining AI models.
Hospitals, health systems, patients, and developers benefit from recommended metrics and benchmarks that improve safety, accountability, and risk assessment for deployed AI systems.
Tech companies and smaller developers face a voluntary framework, so adoption may be limited and inconsistent, reducing near-term usefulness of the standard.
Small developers and research teams face added administrative burden to document models to the new template, increasing costs for startups and academic projects.
Small businesses and developers risk exposing sensitive model details if published, creating IP and security concerns that could discourage transparency.
Based on analysis of 2 sections of legislative text.
Requires NIST to pilot and publish a modular template and technical guidance for documentation accompanying AI models and related data, with stakeholder input and a 12-month report to Congress.
Official title: To direct the National Institute of Standards and Technology to develop best practices and technical guidance on artificial intelligence model documentation, and for other purposes.
Introduced December 4, 2025 by Sarah McBride · Last progress December 4, 2025
Requires NIST to run a pilot program to create a structured, modular template and technical guidance for documentation that should accompany AI models and related data used in public and private sectors. The pilot must solicit public and stakeholder input, coordinate with other agencies and standards organizations, publish the template and guidance online, and report to relevant congressional committees within 12 months with an assessment and plan for permanent implementation if effective. The work is subject to availability of appropriations and is intended to produce voluntary, consensus-based documentation practices (including metadata fields, metrics/benchmarks, and technical guidelines) to improve transparency, interoperability, and evaluation of AI systems.