SB 3308, the Artificial Intelligence Civil Rights Act of 2025, would create a federal civil-rights and consumer-protection framework for “covered algorithms” used in consequential decisions. The bill defines covered algorithms broadly to include machine learning, natural language processing, artificial intelligence, and similar computational systems used in areas such as employment, education, housing, health care, credit, insurance, criminal justice, immigration, government benefits, public accommodations, and elections. It prohibits developers and deployers from using such systems in ways that cause or contribute to disparate impact or other discrimination on the basis of protected characteristics, and it requires them to evaluate and document potential harms before deployment and on an annual basis after deployment.
The bill also imposes extensive transparency, contracting, and governance requirements. Developers and deployers would need to conduct preliminary evaluations, full pre-deployment evaluations when harm is plausible, and post-deployment impact assessments, often with independent auditors. They would have to provide public disclosures, short-form notices, reporting mechanisms, and, in some cases, human alternatives and appeal rights. The bill further restricts off-label use, requires written contracts between developers and deployers, protects whistleblowers, and directs the Federal Trade Commission to issue implementing rules and maintain a public repository of evaluations and assessments. It also authorizes the creation of a federal occupational series for algorithm auditing and additional federal staffing and resources.
If enacted, the bill would significantly expand federal oversight of AI and algorithmic decision systems and would likely preempt or supplement existing compliance practices for companies and public entities using automated tools in high-impact settings. It would make violations of its requirements enforceable as unfair or deceptive acts or practices under the FTC Act, while also authorizing state attorneys general, state data protection authorities, and private individuals to bring enforcement actions. The bill would affect developers, deployers, auditors, and entities using automated systems in commercial or government contexts, and it would require retention and public reporting of many algorithm-related assessments for up to 10 years.
The overall sentiment reflected by the bill’s sponsors is strongly protective of civil rights, privacy, transparency, and human oversight in AI systems. Because there are no committee transcripts or votes provided, there is no recorded debate or formal vote history to indicate broader legislative support or opposition. The bill’s structure suggests a precautionary approach to AI governance, emphasizing discrimination prevention, accountability, and user rights.
Potential points of contention are likely to include the breadth of the definition of covered algorithms and consequential actions, the compliance burden on developers and deployers, the scope of required disclosures and public reporting, and the use of disparate-impact standards in AI regulation. Other likely concerns include the role and authority of the FTC, the requirement for independent audits, the private right of action with treble damages, and the limits on arbitration agreements. Supporters would likely view these provisions as necessary safeguards against algorithmic discrimination and opaque automated decision-making, while critics may argue they are overly expansive or difficult to implement.
The bill would create a new federal statutory regime governing the development, deployment, disclosure, auditing, and enforcement of AI and other computational algorithms used in consequential decisions. It would add new obligations for developers and deployers, authorize FTC rulemaking and enforcement, empower state attorneys general and private plaintiffs, and require public reporting, record retention, and human-review options in certain cases. It would also direct OPM to create a federal algorithm-auditing occupational series and authorize additional federal personnel and appropriations.
The bill’s stated purpose and structure reflect a strong pro-civil-rights, pro-transparency, and pro-accountability sentiment toward AI systems. With no committee transcript or vote record available, there is no documented opposition or bipartisan negotiation in the provided materials. The available context indicates the bill is being advanced as a consumer and civil-rights protection measure rather than a deregulatory or innovation-first proposal.
Likely points of contention include whether the bill’s disparate-impact standard is too broad for AI systems, whether the compliance and audit requirements are too costly or technically difficult, and whether the public disclosure and repository provisions could expose trade secrets or sensitive data. The private right of action, damages provisions, limits on arbitration, and FTC enforcement authority may also draw opposition from industry stakeholders. Supporters are likely to emphasize protections for individuals affected by automated decisions, especially in employment, housing, credit, health care, criminal justice, and elections.