HR6356-119

What This Bill Does

The Artificial Intelligence Civil Rights Act is a broad algorithmic-accountability bill. It defines covered algorithms, developers, deployers, personal data, consequential actions, harms, independent auditors, public accommodations, and other terms across employment, education, housing, utilities, health care, credit, insurance, criminal justice, and related services. Title I bars developers and deployers from offering, licensing, selling, or using covered algorithms in ways that cause disparate impact, discrimination, or denial of equal access, while preserving self-testing, auditing, diversity-expansion, security research, and private-club exceptions. Section 102 requires preliminary harm evaluations before covered algorithms are deployed, licensed, offered, or materially changed for consequential actions; full predeployment evaluations when harm is plausible; postdeployment impact assessments; annual developer reviews; FTC submissions; summaries; independent auditor independence; and retention of documentation. Title II requires developers and deployers to prevent and mitigate identified harms, give independent auditors necessary information, consult impacted communities, certify likely safety and nondiscrimination, share compliance information between developers and deployers, use written contracts for third-party systems, provide FTC-regulated human alternatives for certain consequential actions within two years, and protect people who raise concerns or refuse to waive rights. Title III requires plain-language disclosures, assessment summaries, personal-data and transfer information, individual-rights descriptions, FTC explanation studies, consumer-awareness webpages, reports, guidance, and accessible multilingual materials. Title IV makes violations enforceable by the FTC as unfair or deceptive acts or practices, lets State attorneys general and State data protection authorities sue for injunctions, compliance, civil penalties of $15,000 per violation or 4 percent of average gross annual revenue, damages, restitution, fees, and other relief, and creates a private right of action with treble damages or $15,000 per violation, nominal and punitive damages, attorney fees, FTC and State notice, intervention rights, and bans on predispute arbitration and class-action waivers. Title V directs OPM to create a federal algorithm-auditing occupational series and authorizes necessary appropriations plus up to 500 additional FTC personnel.

Who Benefits and How

Individuals subject to AI decisions benefit because the bill targets algorithmic discrimination in employment, education, housing, utilities, health care, credit, insurance, public accommodations, and other consequential contexts. Workers, tenants, students, patients, borrowers, and consumers benefit from impact assessments, disclosures, human alternatives, retaliation protection, enforcement, and private remedies. Independent AI auditors, civil-rights lawyers, explainability providers, accessibility and translation vendors, and federal algorithm-auditing professionals benefit from new compliance, litigation, and government hiring demand. State attorneys general and data protection authorities gain enforcement tools.

Who Bears the Burden and How

AI developers, deployers, employers, lenders, insurers, landlords, schools, health systems, utilities, telecom carriers, common carriers, and nonprofits using covered algorithms bear substantial compliance costs for evaluations, audits, mitigation, contracts, disclosures, documentation, human alternatives, and litigation risk. The FTC, OPM, State attorneys general, and State data protection authorities must build enforcement, rulemaking, guidance, reporting, and staffing capacity. Federal taxpayers fund added agency work, including up to 500 additional FTC personnel.

Key Provisions

• Defines covered algorithms and consequential actions across employment, education, housing, utilities, health care, credit, insurance, public accommodations, and criminal justice.
• Prohibits covered algorithms that cause disparate impact, discrimination, or denial of equal access outside listed testing and research exceptions.
• Requires predeployment evaluations, postdeployment impact assessments, annual reviews, FTC submissions, independent audits, and record retention.
• Requires developers and deployers to mitigate identified harms, consult impacted communities, certify compliance, and share information through contracts.
• Directs FTC rulemaking on human alternatives for covered algorithmic consequential decisions.
• Requires public disclosures, assessment summaries, consumer-awareness materials, explanation studies, multilingual access, and FTC reports.
• Authorizes FTC, State, and private enforcement with civil penalties, damages, fees, intervention rights, and bans on predispute arbitration and class waivers.
• Creates an OPM algorithm-auditing occupational series and authorizes necessary appropriations plus up to 500 FTC personnel.