READ AI Models Act
- Bill Number
- H.R. 6461
- Origin Chamber
- House
- Congress
- 119th Congress, Session 1
- Policy Area
- Science, Technology, Communications
- Status
- Introduced
- Latest Action
- 2026-06-25: Ordered to be Reported in the Nature of a Substitute (Amended) by the Yeas and Nays: 35 - 0.
- Last Updated
- 2026-07-08T08:05:58Z
AI-Generated Summary
Purpose
The READ AI Models Act (H.R. 6461) aims to promote transparency and accountability in artificial intelligence (AI) by directing the National Institute of Standards and Technology (NIST) to create voluntary resources for documenting AI models and their associated data. These resources are intended for use by both public and private sectors to help users understand key details about AI systems.
Key Provisions
- Pilot Program Establishment: NIST's Director must start a pilot program (subject to funding) to develop a structured template and technical guidelines for AI model documentation. This includes consulting with other federal agencies as needed.
- Template Content: The template is modular (flexible sections that can be adapted for specific uses, like industry needs or audiences) and covers details such as:
- Model name.
- Developer's identity and incorporation location.
- Release date.
- Knowledge cutoff date for training data (the latest point in time the model's data covers).
- Supported languages.
- Terms of service.
- Other relevant information as determined by NIST.
- Technical Guidelines: These accompany the template and include metrics or benchmarks for different AI model types, incorporating voluntary industry standards and best practices.
- Stakeholder Engagement: NIST must collaborate with private companies, universities, nonprofits, international standards groups, and federal agencies through outreach. A draft must be published in the Federal Register for at least 60 days of public comments.
- Reporting and Publication: Within 12 months of starting the pilot, NIST must:
- Report to House and Senate committees on the program's effectiveness and, if successful, a plan for permanent implementation.
- Publish the final template and guidelines on NIST's website, incorporating public feedback.
- Definitions: Key terms include "artificial intelligence" (as defined in existing law, referring to systems that mimic human intelligence), "AI model" (software using computational or machine-learning techniques to generate outputs from inputs), and others like "nonprofit organization" (tax-exempt under IRS rules).
Significant Changes to Existing Law
This bill introduces no direct amendments to prior laws. Instead, it builds on existing frameworks, such as the National Artificial Intelligence Initiative Act of 2020 (which defines AI) and the Higher Education Act (for university definitions), by authorizing NIST to create new voluntary tools without mandating their use or altering regulatory requirements.
Potential Impacts
- Government Agencies: NIST will lead development, potentially streamlining federal AI oversight and procurement by providing standardized documentation tools, which could reduce risks in AI adoption across agencies.
- Citizens: Enhances public understanding and trust in AI systems through better documentation, helping individuals assess AI tools in everyday applications like apps or services, though adoption is voluntary.
- Private Sector: Encourages companies to document AI models consistently, which may lower development costs via shared best practices but could add minor administrative burdens for documentation.
- International Relations: By aligning with global standards organizations, it supports U.S. leadership in AI governance, potentially influencing international norms without binding foreign entities.
Main Stakeholders Affected
- Developers and Users of AI: Private companies, public sector entities, and researchers who build or deploy AI models, benefiting from flexible documentation tools.
- Federal Government: NIST (primary implementer), other agencies (consulted for input), and congressional committees (receiving reports).
- Broader Community: Universities, nonprofits, and international standards groups involved in collaboration and feedback; the public through accessible online resources.
Notable Legal, Constitutional, or Political Implications
- Legal: The bill emphasizes voluntary guidelines, avoiding new regulations or enforcement mechanisms, which limits legal obligations but could inform future mandatory rules if the pilot succeeds.
- Constitutional: No direct implications, as it involves non-binding standards and public-private collaboration without infringing on free speech or due process.
- Political: Positions Congress as proactive on AI ethics and transparency, potentially bridging bipartisan interests in technology policy; success could expand NIST's role in emerging tech without partisan mandates.
This summary was generated by AI and may contain inaccuracies. Refer to the official source document for the authoritative text.
Sponsor
Rep. McBride, Sarah [D-DE-At Large]
Cosponsors (9)
Rep. Obernolte, Jay [R-CA-23], Rep. Lawler, Michael [R-NY-17], Rep. Carson, André [D-IN-7], Rep. Auchincloss, Jake [D-MA-4], Rep. Whitesides, George [D-CA-27], Rep. Gottheimer, Josh [D-NJ-5], Rep. Scholten, Hillary J. [D-MI-3], Rep. Crockett, Jasmine [D-TX-30], Rep. Kean, Thomas H. [R-NJ-7]
Recent Actions
- 2026-06-25: Ordered to be Reported in the Nature of a Substitute (Amended) by the Yeas and Nays: 35 - 0.
- 2026-06-25: Committee Consideration and Mark-up Session Held
- 2025-12-04: Referred to the House Committee on Science, Space, and Technology.
- 2025-12-04: Introduced in House
- 2025-12-04: Introduced in House
Bill Versions
- Resources for Evaluating and Documenting AI Models — issued 2025-12-04 — PDF (6 pages)