Artificial Intelligence Civil Rights Act of 2025
- Bill Number
- H.R. 6356
- Origin Chamber
- House
- Congress
- 119th Congress, Session 1
- Policy Area
- Science, Technology, Communications
- Status
- Introduced
- Latest Action
- 2025-12-02: Referred to the Committee on Energy and Commerce, and in addition to the Committee on Oversight and Government Reform, for a period to be subsequently determined by the Speaker, in each case for consideration of such provisions as fall within the jurisdiction of the committee concerned.
- Last Updated
- 2026-04-30T08:06:36Z
AI-Generated Summary
Artificial Intelligence Civil Rights Act of 2025 (H.R. 6356)
Purpose
The legislation aims to protect individual civil rights by regulating the development and use of advanced computational algorithms—such as those powered by artificial intelligence (AI), machine learning, or similar technologies—that influence important life decisions. It seeks to prevent discrimination, harm, and unfair practices in areas like employment, housing, healthcare, and voting, ensuring algorithms do not disproportionately affect people based on traits like race, gender, or disability. The Act builds on existing civil rights laws to address biases in automated systems.
Key Provisions
The bill is structured into five titles, outlining prohibitions, requirements, transparency, enforcement, and federal support.
Title I: Civil Rights
- Prohibition on Discrimination (Sec. 101): Developers (those who create or modify algorithms) and deployers (those who use them commercially) cannot use covered algorithms in ways that cause or contribute to unfair differences in access to opportunities (known as disparate impact) or direct discrimination based on protected characteristics (e.g., race, sex, age, disability, or immigration status). This applies to consequential actions—decisions with significant effects, such as hiring, loan approvals, or criminal justice outcomes.
- Evaluations and Assessments (Sec. 102):
- Pre-Deployment Evaluations: Before using an algorithm, developers and deployers must assess the risk of harm. If harm is plausible, an independent auditor (an unbiased expert) conducts a detailed review of the algorithm's design, training data, testing, and potential biases. Reports must be submitted to the Federal Trade Commission (FTC) and summarized publicly.
- Post-Deployment Impact Assessments: Deployers must annually check for real-world harms and submit findings to the FTC if issues arise.
- Developers review deployers' assessments yearly to monitor performance and compliance.
- All records must be kept for 10 years, with protections for trade secrets and personal privacy.
Title II: Covered Algorithm and Contract Standards
- Algorithm Standards (Sec. 201): Requires measures to mitigate identified risks, stakeholder consultations (e.g., affected communities), certifications of safety, and ensuring algorithms perform reliably without deception.
- Developer-Deployer Relationships (Sec. 202): Developers must share evaluation reports and cooperate on audits. Contracts between them must detail data handling and prohibit unauthorized data mixing.
- Human Alternatives and Protections (Sec. 203): The FTC will set rules for individuals to opt out of algorithm-based decisions and request human review (e.g., appealing an AI hiring denial). Algorithms cannot manipulate user choices.
- Anti-Retaliation (Sec. 204): Protects individuals from punishment for raising concerns or exercising rights under the Act; includes whistleblower safeguards for employees reporting violations.
Title III: Transparency
- Notice and Disclosure (Sec. 301): Developers and deployers must publicly post clear disclosures about data practices, rights, and evaluations in multiple languages (top 10 U.S. languages) and accessible formats for people with disabilities. Short-form notices inform users at first interaction.
- Study on Explanations (Sec. 302): The FTC will study requiring explanations of how algorithms affect individuals (e.g., why a loan was denied) and report to Congress.
- Consumer Awareness (Sec. 303): The FTC must create a public webpage explaining the Act, issue annual reports on assessments, and build an online repository of evaluations for public access.
Title IV: Enforcement
- FTC Enforcement (Sec. 401): Violations are treated as unfair or deceptive practices under FTC rules, with expanded jurisdiction over nonprofits, banks, and carriers. The FTC can issue rules and penalties.
- State Enforcement (Sec. 402): State attorneys general can sue for violations, seeking injunctions, penalties (up to $15,000 per violation or 4% of revenue), and damages. They must notify the FTC, which can intervene.
- Private Right of Action (Sec. 403): Individuals can sue for violations, potentially winning treble damages (three times actual harm), punitive damages, or fees. Pre-dispute arbitration clauses are invalid for these claims. Individuals must notify the FTC and state AG first, who can intervene.
- Severability and Construction (Secs. 404-405): Invalid parts do not affect the whole; the Act does not override labor, health, or other laws.
Title V: Federal Resources
- Auditing Positions (Sec. 501): Creates new federal job categories for AI auditing experts.
- Funding (Sec. 502): Authorizes appropriations and allows the FTC to hire up to 500 additional staff.
Significant Changes to Existing Law
- Extends civil rights protections (e.g., from the Civil Rights Act of 1964) to AI systems, introducing disparate impact liability for algorithms—meaning even unintentional biases can lead to violations if not justified by a legitimate need.
- Mandates proactive testing and auditing, unlike prior laws that focused on post-harm remedies.
- Grants individuals new rights to human alternatives and appeals, and voids forced arbitration for AI disputes.
- Expands FTC authority beyond for-profit entities and coordinates with states, filling gaps in current privacy and anti-discrimination frameworks like the Fair Credit Reporting Act or Equal Credit Opportunity Act.
Potential Impacts
- Government Agencies: The FTC gains significant oversight, requiring new resources for rulemaking, investigations, and a public repository. Other agencies (e.g., for labor or health) must align with these standards without reduced authority. States may see increased enforcement roles, potentially straining resources but enhancing local protections.
- Citizens: Improves fairness in AI-driven decisions, reducing biases in hiring, lending, or policing. Provides tools like opt-outs and appeals, empowering individuals—especially from marginalized groups—to challenge harms. However, it may slow AI adoption in some services, affecting efficiency in areas like benefits administration.
- International Relations: Could influence global AI standards by setting U.S. benchmarks for ethical deployment, potentially pressuring foreign tech firms operating in the U.S. No direct international provisions, but data transfer rules may complicate cross-border dealings.
Main Stakeholders Affected
- Developers and Deployers: Tech companies, AI firms, and users of algorithms (e.g., banks, employers, landlords) face compliance costs for audits, disclosures, and mitigations.
- Individuals: U.S. residents benefit from anti-discrimination protections, especially those with protected characteristics; vulnerable groups (e.g., low-income, minorities) gain from bias prevention.
- Government Entities: FTC leads enforcement; states and agencies like the Office of Personnel Management implement auditing and coordination.
- Auditors and Experts: Independent professionals and federal auditors will see new demand for unbiased AI evaluations.
- Communities and Advocacy Groups: Involved in consultations, they can report issues and access public data for accountability.
Notable Legal, Constitutional, or Political Implications
- Legal: Strengthens enforcement through private lawsuits and state actions, potentially increasing litigation over AI biases. The disparate impact standard presumes algorithms are holistic (not separable parts), easing proof of violations but allowing rebuttals. Whistleblower protections align with labor laws, but contracts must not hinder reporting.
- Constitutional: Protects rights under the U.S. Constitution (e.g., due process in consequential actions like bail or voting) by curbing coercive or harmful AI uses. May raise First Amendment questions on disclosures or explanations, though focused on commercial practices.
- Political: Advances equity in emerging tech, reflecting bipartisan concerns on AI ethics (introduced by progressive Democrats). Could spark debates on innovation vs. regulation, with potential for industry pushback on costs. Promotes transparency via public repositories, aiding oversight but risking privacy if not redacted properly.
This summary was generated by AI and may contain inaccuracies. Refer to the official source document for the authoritative text.
Sponsor
Rep. Clarke, Yvette D. [D-NY-9]
Cosponsors (28)
Rep. Lee, Summer L. [D-PA-12], Rep. Pressley, Ayanna [D-MA-7], Rep. Jayapal, Pramila [D-WA-7], Rep. Bell, Wesley [D-MO-1], Rep. Carson, André [D-IN-7], Rep. Chu, Judy [D-CA-28], Rep. Davis, Danny K. [D-IL-7], Rep. Deluzio, Christopher R. [D-PA-17], Rep. Jackson, Jonathan L. [D-IL-1], Rep. Kelly, Robin L. [D-IL-2], Rep. McGovern, James P. [D-MA-2], Del. Norton, Eleanor Holmes [D-DC-At Large], Rep. Ocasio-Cortez, Alexandria [D-NY-14], Rep. Pocan, Mark [D-WI-2], Rep. Ramirez, Delia C. [D-IL-3], Rep. Thompson, Bennie G. [D-MS-2], Rep. Tlaib, Rashida [D-MI-12], Rep. Watson Coleman, Bonnie [D-NJ-12], Rep. Wilson, Frederica S. [D-FL-24], Rep. García, Jesús G. "Chuy" [D-IL-4], Rep. Foushee, Valerie P. [D-NC-4], Rep. Brown, Shontel M. [D-OH-11], Rep. Mfume, Kweisi [D-MD-7], Rep. Hayes, Jahana [D-CT-5], Rep. DeSaulnier, Mark [D-CA-10], Rep. Green, Al [D-TX-9], Rep. Grijalva, Adelita S. [D-AZ-7], Rep. McClellan, Jennifer L. [D-VA-4]
Recent Actions
- 2025-12-02: Referred to the Committee on Energy and Commerce, and in addition to the Committee on Oversight and Government Reform, for a period to be subsequently determined by the Speaker, in each case for consideration of such provisions as fall within the jurisdiction of the committee concerned.
- 2025-12-02: Referred to the Committee on Energy and Commerce, and in addition to the Committee on Oversight and Government Reform, for a period to be subsequently determined by the Speaker, in each case for consideration of such provisions as fall within the jurisdiction of the committee concerned.
- 2025-12-02: Introduced in House
- 2025-12-02: Introduced in House
Bill Versions
- Artificial Intelligence Civil Rights Act of 2025 — issued 2025-12-02 — PDF (72 pages)