Sectoral AI Governance Act of 2026

Applies To

Overview

H.R.9125, introduced in the 119th Congress, would grant federal agencies explicit statutory authority to promulgate rules targeting algorithmic systems that may contribute to violations of the federal statutes those agencies oversee. Rather than establishing a single horizontal AI regulator, the bill adopts a sectoral model: existing bodies such as the CFPB, EEOC, HHS, and others would each develop AI-specific rules within their current jurisdictional mandates. Key provisions would likely require deployers to assess whether their algorithmic decision-making systems pose material legal risk under applicable federal law and to maintain documentation supporting that assessment. Enforcement would follow each agency's existing mechanisms, including civil penalties, corrective orders, and referral to the Department of Justice. The bill was introduced in March 2026 and remains in early legislative review, with no committee markup date publicly scheduled as of the entry date.

Key Requirements

• •Federal agencies are authorized to issue binding rules for algorithmic decision-making systems that may materially contribute to violations of laws they enforce
• •Regulated deployers must assess AI systems for material legal risk under applicable sector-specific federal statutes
• •Documentation of algorithmic system design, decision logic, and risk assessments must be maintained and available to regulators upon request
• •Agency-specific rulemaking timelines and penalty structures will vary by sector; existing civil penalty frameworks apply
• •Organizations operating across multiple regulated sectors may face overlapping obligations from different agencies simultaneously
• •No single compliance threshold is established; materiality determinations are made agency by agency

What Your Organization Must Do

• →Map all algorithmic decision-making systems in production against each federal regulatory regime your organization operates under to identify potential material-risk exposures
• →Engage sector-specific regulatory counsel now to anticipate how agencies such as the CFPB, EEOC, and HHS are likely to frame rulemaking under this authority
• →Build a cross-functional AI governance committee that includes legal, compliance, and technology leads capable of responding to multiple simultaneous agency rulemakings
• →Update AI system documentation practices to capture decision logic, training data provenance, and risk assessments in a format suitable for regulatory examination
• →Revise vendor and third-party AI procurement agreements to require disclosure of algorithmic decision-making components and supporting compliance documentation
• →Monitor Congressional progress and agency regulatory agendas closely to set realistic internal deadlines for compliance program updates

Playbook Guidance

Step-by-step implementation guidance for compliance teams.

Frequently Asked Questions

Which federal agencies would have rulemaking authority under SAIGA 2026?

The bill explicitly contemplates agencies including the CFPB, EEOC, and HHS, but authority extends to any federal agency with existing statutory enforcement jurisdiction. Each agency would develop its own AI-specific rules within its current mandate, meaning the list of relevant regulators depends entirely on the sectors where your organization operates.

Does SAIGA 2026 apply to AI vendors or only to organizations that deploy AI systems?

The bill targets deployers whose algorithmic decision-making systems could materially contribute to violations of federal law in regulated sectors. Vendors are not the primary regulated party, but deployers will likely need to flow down documentation and disclosure obligations contractually to third-party AI suppliers to satisfy regulatory examination requirements.

What does 'material legal risk' mean under SAIGA 2026 and who determines it?

The bill does not establish a single universal materiality standard. Each agency makes its own determination based on the federal statutes it enforces, which means a CFPB materiality assessment for a credit-scoring algorithm could differ substantially from an EEOC assessment of the same system used in hiring.

What civil penalties could apply for non-compliance with SAIGA 2026?

No new penalty structure is created by the bill itself. Enforcement follows each agency's existing civil penalty frameworks, so exposure varies significantly by sector. CFPB penalties, for example, can reach tens of thousands of dollars per day per violation, while other agencies operate under different statutory caps.

How does SAIGA 2026 differ from the EU AI Act in its compliance approach?

Unlike the EU AI Act's horizontal risk-tier classification system administered by a centralized authority, SAIGA 2026 uses a sectoral model that delegates rulemaking to existing US agencies. There is no single compliance framework or unified enforcement body, so multi-sector organizations must track and satisfy potentially simultaneous, divergent agency rulemakings.

What should compliance teams do now given SAIGA 2026 has no scheduled committee markup?

Organizations should treat the absence of a markup date as preparation time rather than a reason to delay. Mapping algorithmic systems against applicable federal regulatory regimes, updating documentation practices, and engaging sector-specific counsel now positions compliance programs to respond quickly once agency rulemaking timelines become clearer.