Eliminating Bias in Algorithmic Systems Act of 2026
- Bill Number
- S. 3680
- Origin Chamber
- Senate
- Congress
- 119th Congress, Session 2
- Policy Area
- Civil Rights and Liberties, Minority Issues
- Status
- Introduced
- Latest Action
- 2026-01-15: Read twice and referred to the Committee on Homeland Security and Governmental Affairs.
- Last Updated
- 2026-02-11T17:25:37Z
AI-Generated Summary
Purpose
The Eliminating Bias in Algorithmic Systems Act of 2026 aims to address potential biases, discrimination, and other harms caused by algorithms (automated decision-making tools like artificial intelligence or machine learning systems) used by federal agencies. It requires agencies to establish dedicated civil rights offices to monitor and mitigate these issues, particularly those affecting protected characteristics such as race, sex, disability, or age.
Key Provisions
- Definitions:
- Agency: Refers to any federal executive department or independent agency, as defined under existing U.S. law (44 U.S.C. § 3502).
- Covered agency: Any agency that uses, funds, develops, procures, oversees, regulates, or advises on "covered algorithms."
- Covered algorithm: A computational process (e.g., using machine learning, AI, or similar techniques) that could significantly impact access to government programs, regulated economic opportunities (like loans or jobs), or protected rights.
- Protected characteristic: A broad list of traits including race, ethnicity, sex (encompassing sexual orientation and gender identity), disability, age, religion, income level, and any other category safeguarded by federal anti-discrimination laws.
- Establishment of Civil Rights Offices (Section 3(a)):
- Each covered agency must maintain or create an office of civil rights staffed with experts and technologists.
- These offices focus on identifying and addressing biases, discrimination, or harms in algorithms that target or disproportionately affect individuals or groups based on protected characteristics.
- Reporting Requirements (Section 3(b)):
- Within one year of enactment, and every two years thereafter, each office must submit reports to relevant congressional committees.
- Reports cover: (1) the current state of algorithmic technology and risks of bias/discrimination within the agency's scope; (2) steps taken to reduce harms; (3) engagement with stakeholders like civil rights groups, industry, academics, workers, and affected communities; and (4) recommendations for new laws or agency actions.
- Interagency Working Group (Section 3(c)):
- The Department of Justice's Civil Rights Division must form a working group within one year, including representatives from each covered agency's civil rights office, to coordinate efforts on algorithmic civil rights issues.
- Funding (Section 3(d)):
- Authorizes Congress to appropriate necessary funds for agencies to implement the Act.
Significant Changes to Existing Law
- This Act introduces a new mandate for federal agencies to proactively address algorithmic bias, which was not previously required under general civil rights laws (e.g., Title VI of the Civil Rights Act of 1964 or the Administrative Procedure Act).
- It expands oversight by requiring specialized offices and biennial reporting, building on but going beyond existing agency civil rights offices that handle broader discrimination complaints.
- No explicit amendments to prior statutes, but it integrates algorithmic harms into civil rights enforcement frameworks.
Potential Impacts
- Government Agencies: Covered agencies (potentially dozens, including those in health, finance, housing, and justice) will face new administrative burdens, such as hiring specialists, conducting audits, and producing reports. This could increase operational costs but improve accountability in AI use.
- Citizens: Individuals and communities with protected characteristics may benefit from reduced discriminatory outcomes in government services (e.g., fairer AI-driven benefit eligibility or loan approvals), enhancing trust in federal programs. However, it does not create private rights to sue agencies directly.
- International Relations: Minimal direct impact, as the Act focuses on domestic federal agencies; it could indirectly influence U.S. standards for AI ethics in global tech partnerships or aid programs.
Main Stakeholders Affected
- Federal Agencies: Primary implementers, especially those deploying AI (e.g., Department of Health and Human Services, Department of Housing and Urban Development, or Federal Trade Commission).
- Congress and Oversight Bodies: Recipients of reports, with jurisdiction over agency activities.
- Civil Society and Experts: Civil rights organizations, consumer advocates, academics, tech workers, and affected populations (e.g., low-income or minority groups) involved in consultations and recommendations.
- Industry and Businesses: Tech firms developing or selling algorithms to the government, who may need to adapt to bias-mitigation requirements in contracts.
Notable Legal, Constitutional, or Political Implications
- Legal: Strengthens enforcement of existing anti-discrimination laws by applying them to emerging AI technologies; the broad definition of "covered algorithm" could lead to challenges over what qualifies as a "material effect," potentially requiring court clarification.
- Constitutional: Aligns with Equal Protection Clause principles (under the 14th Amendment) by targeting disparate impacts on protected groups, but avoids mandating specific outcomes to prevent overreach into agency discretion.
- Political: Represents a bipartisan push (introduced by Senators Markey, Merkley, et al.) to regulate AI ethically amid growing concerns over tech biases; it could spark debates on government overregulation of innovation versus protecting vulnerable populations, influencing future AI legislation like comprehensive federal AI bills.
This summary was generated by AI and may contain inaccuracies. Refer to the official source document for the authoritative text.
Sponsor
Cosponsors (6)
Sen. Merkley, Jeff [D-OR], Sen. Welch, Peter [D-VT], Sen. Warren, Elizabeth [D-MA], Sen. Schiff, Adam B. [D-CA], Sen. Wyden, Ron [D-OR], Sen. Booker, Cory A. [D-NJ]
Recent Actions
- 2026-01-15: Read twice and referred to the Committee on Homeland Security and Governmental Affairs.
- 2026-01-15: Introduced in Senate
Bill Versions
- Eliminating Bias in Algorithmic Systems Act of 2026 — issued 2026-01-15 — PDF (6 pages)