HR6356-119
In CommitteeArtificial Intelligence Civil Rights Act of 2025
Judy Chu’s recorded connection
Cosponsored this bill
-
Potential burden on Covered algorithm developers
More compliance requirements
This section requires AI developers and deployers to publish detailed public disclosures about their AI practices, including contact information, data practices, audit results, and how individuals can exercise their rights. Notices must be available in the 10 most common US languages and accessible to people with disabilities. Short-form notices (under 500 words) must be provided when individuals first interact with AI systems. Companies face significant disclosure requirements, while translation services and accessibility consultants will see new demand.
-
Potential burden on Covered algorithm developers
Higher risk
This section authorizes state attorneys general and state data protection authorities to sue companies that violate the AI Civil Rights Act. States can seek penalties of $15,000 per violation OR 4% of the company gross annual revenue (whichever is higher), plus damages, attorneys fees, and court orders. States must notify the FTC before filing suit, and the FTC can intervene in state actions. Companies face enforcement risk from 50+ state-level enforcers in addition to federal authorities.
-
Potential burden on Covered algorithm developers
More compliance requirements
This provision requires AI developers and deployers to evaluate their algorithms for potential harm before deployment and conduct annual impact assessments afterward. If harm is plausible, companies must hire independent auditors to conduct detailed reviews. Results must be submitted to the FTC, published publicly in summary form, and retained for 10 years. AI auditing firms will see major new business opportunities, while tech companies and businesses face significant compliance costs for documentation, testing, and auditing.
-
Unclear effects on Covered algorithm developers
More compliance requirements
This section defines key terms for the entire AI Civil Rights Act, including what counts as a "covered algorithm" (AI systems using machine learning for important decisions), what "consequential actions" the law covers (employment, housing, credit, healthcare, insurance, education, criminal justice, elections, and government benefits), and what "protected characteristics" are shielded from discrimination (race, sex, disability, age, religion, income level, and others). AI developers and companies using AI systems will need to understand these definitions to determine if they are subject to the law.
-
Potential burden on Covered algorithm developers
More compliance requirements
This provision prohibits AI developers and companies from offering or using AI algorithms that discriminate against protected groups or cause unjustified disparate impact in access to jobs, housing, credit, or other important opportunities. Individuals from protected classes (racial minorities, women, people with disabilities, etc.) benefit from legal protection against algorithmic bias. Companies using AI for hiring, lending, or insurance face liability if their algorithms have discriminatory effects. Limited exceptions exist for bias testing, diversity efforts, security research, and private clubs.
-
Potential burden on Covered algorithm developers
More compliance requirements
This section establishes safety and performance standards for AI systems, requiring companies to prevent harm, consult with affected communities, and certify that their algorithms will not cause discrimination or deception. Companies must ensure AI performs reasonably and matches advertised capabilities. It is illegal to market AI deceptively or use it for purposes not evaluated in pre-deployment testing. Tech companies and AI users face compliance burdens, while AI testing and certification firms benefit from new business.
-
Potential burden on Covered algorithm developers
Higher risk
This provision grants the Federal Trade Commission authority to enforce the AI Civil Rights Act, treating violations as unfair or deceptive practices. Importantly, it extends FTC jurisdiction to entities typically exempt from FTC oversight, including non-profit organizations, banks, credit unions, airlines, telecommunications carriers, and agricultural businesses. All of these sectors now face FTC enforcement for AI discrimination, expanding the law reach significantly beyond typical consumer protection cases.
-
Potential burden on Covered algorithm developers
Higher risk
This provision creates a private right of action allowing individuals harmed by AI discrimination to sue directly in court. Plaintiffs can recover treble damages or $15,000 per violation (whichever is greater), plus punitive damages and attorneys fees. The section bans pre-dispute arbitration agreements and class action waivers, meaning companies cannot force AI discrimination claims into private arbitration. Plaintiffs attorneys and civil rights lawyers will see major new litigation opportunities, while companies face significant exposure to class action lawsuits.
-
Potential burden on Covered algorithm developers
More compliance requirements
This provision establishes rules for business relationships between AI developers (who create algorithms) and deployers (who use them). Developers must share compliance documentation, allow auditor access, and use detailed written contracts specifying data handling, deployment instructions, and notification of algorithm changes. Contracts must be retained for 10 years. Tech companies and their business customers face new contractual requirements, while technology contract lawyers will see increased demand for their services.
-
Potential burden on Covered algorithm developers
More compliance requirements
This provision prohibits companies from retaliating against individuals who exercise their rights under the AI Civil Rights Act or who report violations as whistleblowers. Employees cannot be fired, demoted, or harassed for raising concerns about AI discrimination or cooperating with investigations. Individuals and employees gain protection when challenging AI systems, while companies face restrictions on how they can respond to complaints and internal concerns.