WeVote
Log In

Bill

Bill

HR 9125

Sectoral AI Governance Act of 2026

119th Congress Introduced by Valerie Foushee and 2 co-sponsors

Gives agencies power to regulate AI-based decision systems likely to cause federal-law violations, with cross-agency coordination, risk mitigation, and regular reviews.

Introduced in House
0
WeVote Research Nonpartisan
Bill Summary · HR 9125

Overview

HR 9125, the Sectoral AI Governance Act of 2026, would authorize heads of federal agencies to issue prospective rules regulating the use of algorithmic decision-making systems (ADMS) that are likely to materially contribute to violations of federal law the agency enforces. The act aims to create a coordinated, cross-agency framework for regulating ADMS while preserving agency enforcement authority and promoting transparency, coordination, and periodic review.

Purpose and intent

  • Provide clear statutory authority for agencies to regulate ADMS that pose a material risk of violating federal law.
  • Improve consistency and coordination across agencies through shared guidance, consultation, and reporting.
  • Mitigate disruptive effects on government services and public benefits when regulating ADMS.
  • Establish ongoing assessment, reporting, and periodic review to ensure rules remain appropriate as technology evolves.

Key provisions and changes

  • Section 3: Rulemaking Authority Related to the Use of ADMS
    • Agencies authorized to enforce federal law may issue rules to regulate ADMS if the agency determines, based on available evidence, that the system is likely to materially contribute to violations and to mitigate those violations.
    • Violations of these agency rules would be enforceable as violations of the federal laws the agency enforces.
    • Advanced notice of proposed rulemaking (ANPRM) required at least 60 days before proposing a rule, describing the targeted ADMS, implicated laws, and information sought, with exceptions if prior rulemaking records justify bypassing ANPRM.
  • Section 3(b): Consultation and Coordination During Rulemaking
    • Preliminary consultation with the Office of Information and Regulatory Affairs (OIRA) to identify overlaps or conflicts with other agencies’ proposed rules.
    • Mandatory technical consultation with the Office of Science and Technology Policy (OSTP) and National Institute of Standards and Technology (NIST) on technical characteristics, governance, and evaluation of ADMS.
    • Efforts to coordinate across agencies to avoid conflicting requirements.
  • Section 3(c): Mitigation of Disruption to Benefits
    • Agencies must consider impacts on government services or public benefits, and seek to minimize adverse effects on delivery, accessibility, timeliness, and integrity where feasible.
  • Section 3(d): Guidance
    • OMB, through OIRA, with OSTP input, to issue guidance to resolve conflicts and ensure cross-agency consistency.
  • Section 3(e): Periodic Review
    • Rules must be reviewed at least every 3 years after promulgation (and not less than every 5 years thereafter) to determine continued appropriateness and tailoring.
    • Agencies must repeal or amend rules that no longer properly mitigate violations.
  • Section 3(f): Biennial Reporting
    • Agencies that enacted ADMS rules must report every two years on rulemaking details, including:
    • Description of rulemaking and the ADMS involved.
    • Enforcement actions, penalties, remedies, and categories of violations.
    • Staffing efforts and resource needs.
    • Challenges and disparities or discriminatory effects considered.
    • How the rule regulates ADMS to prevent violations.
    • Any overlap or conflicts with other agency rules and how resolved.
    • Agencies that did not engage in rulemaking must report on ADMS that may materially contribute to violations and explain reasons for not rulemaking, plus any alternative mitigation approaches.
  • Definitions
    • ADMS: Computational processes in software/hardware using statistics, ML/AI, or other data processing that make or support decisions and can alter outcomes (non-passive infrastructure).
    • Authority to Enforce: Power to impose sanctions or pursue federal court actions to remedy violations.
    • Materially Contribute: Plays a significant role in causing or enabling violations.
    • AI: As defined in the National AI Initiative Act of 2020.
    • Passive Computing Infrastructure: Intermediary tech that does not influence decisions (e.g., hosting, routing, storage).

Who would be affected

  • Federal agencies with enforcement authority on federal laws potentially impacted by ADMS.
  • Agencies’ staff responsible for rulemaking, enforcement, and compliance relating to algorithmic systems.
  • Contractors or service providers involved in public benefits or government services using ADMS.
  • Stakeholders providing input during ANPRM and rulemaking processes, including technologists, researchers, and affected communities.

Procedural and timeline aspects

  • Rulemaking under this act would proceed under 5 U.S.C. 553 (notice-and-comment) with an ANPRM 60 days before the proposed rule, unless an exception applies.
  • Inter-agency coordination and technical consultations are required during rulemaking.
  • Periodic rule reviews are mandated at least every 3 years after adoption (and not less than every 5 years).
  • Biennial reporting begins within two years of enactment for agencies that enacted rules, with subsequent reports every two years; non-enacting agencies must still report on ADMSs identified as likely to cause violations and rationale for not rulemaking.

Notes

  • The bill explicitly states non-preemption of state authority unless conflicts arise with federal law or agency rules.
  • Co-sponsors include Valerie Foushee and Sara Jacobs.

Compiled from official sources — confirm details with the bill’s official record.

Sign in to ask a question.