Sectoral AI Governance Act of 2026
- Sponsor
- Rep. Jacobs, Sara [D-CA-51]
- Committees
- Oversight and Government Reform Committee (primary)
- Last action
- Jun 3, 2026
Bottom line
This bill aims to provide a clear framework for federal agencies to proactively regulate harmful uses of AI systems within their existing enforcement mandates, promoting consistency and accountability across the government.
What it actually does
This bill authorizes federal agencies to issue rules regulating algorithmic decision-making systems if they are likely to materially contribute to violations of federal laws the agency enforces. It establishes procedures for public comment, inter-agency consultation, and periodic review of these rules. The bill also mandates biennial reports from agencies on their rulemaking activities or reasons for not engaging in rulemaking related to AI systems.
Proponents argue
Supporters argue that this bill addresses a critical gap in federal oversight of artificial intelligence by empowering existing agencies to regulate AI within their specific domains. They contend it provides necessary clarity and a coordinated framework, preventing a patchwork of inconsistent regulations while allowing agencies to mitigate potential harms from AI systems that could violate civil rights, consumer protection, or other federal laws.
Opponents contend
Critics might argue that this bill could lead to an overreach of regulatory power by granting broad rulemaking authority to numerous agencies, potentially stifling innovation in AI development. They may also raise concerns about the potential for inconsistent application across agencies despite coordination efforts, or that the definition of 'algorithmic decision-making system' is too broad, capturing technologies not intended for regulation.
The bill is relatively short and clearly structured, making it accessible for an informed reader to understand its core provisions within a reasonable timeframe.
Section 3(a)(1)
Agency Rulemaking Authority for Algorithmic Systems
This provision grants federal agencies the power to create new rules specifically to regulate algorithmic decision-making systems. This authority is triggered when an agency determines, based on available evidence, that an algorithmic system is likely to significantly contribute to violations of a federal law that the agency is already authorized to enforce. The rules are intended to prevent these violations.
Supporters argue
Supporters argue this provision is crucial for adapting existing federal laws to the challenges posed by rapidly evolving AI technologies. It allows agencies with subject-matter expertise to proactively address AI-related harms within their specific sectors (e.g., FTC for consumer protection, EEOC for civil rights) without needing entirely new legislation for each AI application.
Critics contend
Opponents might express concern that this grants overly broad powers to agencies, potentially leading to a fragmented and inconsistent regulatory landscape across different sectors, even with coordination efforts. They might also argue it could create regulatory uncertainty that stifles innovation.
Tradeoffs
Balances the need for adaptive regulation of emerging technologies with concerns about potential regulatory overreach and fragmentation.
Section 3(a)(2)
Algorithmic Rule Violations as Federal Law Violations
This provision clarifies that if an agency issues a rule under this Act and that rule is violated, the violation will be treated as if it were a violation of the underlying federal law that the agency is authorized to enforce. This means agencies can use their existing administrative and civil enforcement powers to address non-compliance with these new AI-specific rules.
Section 3(a)(3)
Public Comment on Algorithmic System Harms
Before an agency can issue a formal notice of proposed rulemaking for an AI-related rule, it generally must first publish an 'advanced notice of proposed rulemaking' (ANPRM) at least 60 days prior. This ANPRM is designed to gather public comment specifically on whether the algorithmic system in question is indeed likely to materially contribute to federal law violations. It requires the agency to describe the system, implicated laws, and types of information sought.
Section 3(b)
Cross-Agency AI Regulatory Harmonization
This section mandates that agencies consult with the Office of Information and Regulatory Affairs (OIRA) to identify and resolve potential overlaps or conflicts with other agencies' proposed AI rules. They must also consult with the Office of Science and Technology Policy (OSTP) and the National Institute of Standards and Technology (NIST) on the technical aspects of the AI systems they seek to regulate. Agencies are also required to coordinate to prevent conflicting requirements.
Section 3(c)
Protecting Government Services and Public Benefits
When an agency creates a rule under this Act, it must consider whether the algorithmic system being regulated is used in connection with government services or public benefits (including by contractors). Where practical, the agency must try to minimize any unnecessary negative effects of the rule on the delivery, accessibility, timeliness, integrity, or continuity of these services or benefits.
Section 3(e)
Adaptive Review of AI Regulations
This provision requires agencies to review any AI-related rules they promulgate. The first review must occur within three years of the rule's issuance, and subsequent reviews at least every five years. The purpose is to assess if the rule remains appropriate and effectively tailored to mitigate violations, especially considering technological developments or changes in how algorithmic systems are used. Agencies must amend or repeal rules found to be inappropriate or not well-tailored.
Section 3(f)
Transparency and Accountability in AI Governance
This section mandates biennial reports to Congress, OMB, and OSTP. Agencies that *have* engaged in rulemaking must report on their rules, the systems mitigated, staffing, challenges, resources needed, enforcement actions, how rules prevent violations, consideration of disparate impacts, and how inter-agency conflicts were addressed. Agencies that *identified* potential violations but *did not* engage in rulemaking must report on the identified uses, reasons for not rulemaking, and alternative mitigation mechanisms used.
Defines 'algorithmic decision-making system' broadly to include computational processes based on statistics, machine learning, AI, or other data processing techniques that make, execute, or assist in decisions capable of altering outcomes, specifically excluding 'passive computing infrastructure.'
Section 3(g)(1)
Why it matters:Definitions are standard legislative practice, but the precise scope of 'algorithmic decision-making system' is central to the bill's applicability and could be easily overlooked. The exclusion of 'passive computing infrastructure' is a key carve-out that limits the bill's reach.
Case for: Proponents argue that a broad definition is necessary to ensure the bill's relevance as AI technology evolves, preventing it from becoming quickly outdated. The exclusion of 'passive computing infrastructure' is intended to avoid regulating basic internet infrastructure, focusing on systems that actively influence decisions.
Case against: Critics might argue the definition is still too broad and could inadvertently capture technologies not intended for regulation, leading to overreach. Conversely, some might argue the 'passive computing infrastructure' exclusion could be exploited or interpreted too broadly, creating loopholes.
Estimated impact: This definition fundamentally determines the scope of all agency rulemaking under the Act, impacting potentially every sector where AI is used to make or assist decisions.
Defines 'materially contribute' as playing a significant role in bringing about or enabling a violation of Federal law, including assisting or facilitating, but explicitly *not* including minimal or incidental support unrelated to the violation.
Section 3(g)(4)
Why it matters:Standard legislative practice for defining terms, but this particular definition sets a critical standard for when agencies can act.
Case for: Proponents argue this definition provides a necessary threshold, ensuring agencies focus on AI systems that genuinely pose a risk of contributing to legal violations, rather than those with negligible or indirect involvement. It aims to prevent overreach while still enabling effective regulation.
Case against: Critics might argue that 'significant role' could still be subject to broad interpretation, leading to inconsistent application across agencies. Conversely, some might argue it sets too high a bar, potentially allowing AI systems that contribute to violations in a less direct but still impactful way to escape regulation.
Estimated impact: This definition directly influences an agency's ability to determine if an algorithmic system warrants regulation, thus shaping the entire scope of the Act's application.