Senate

Led by Senator Edward Markey (D-Mass.) and Representative Yvette Clarke (D-N.Y.), the AI Civil Rights Act would require audits, transparency, and human oversight of AI-driven decisions that affect civil rights.

Democratic lawmakers reintroduced the AI Civil Rights Act, a proposal to limit discriminatory outcomes from artificial intelligence systems used in areas such as hiring, housing, lending, healthcare, and education. The bill seeks to apply civil rights protections to automated decision-making tools that affect access to essential services and opportunities.

At a press conference announcing the bill’s return, Senator Ed Markey said companies have increasingly relied on AI systems to make consequential decisions about people’s lives. Supporters of the legislation argue that without clear safeguards, automated systems can replicate or amplify existing biases, particularly when trained on historical data that reflects past discrimination.

Under the bill, developers and users of AI systems would be prohibited from deploying tools that discriminate based on protected characteristics or that produce unjustified disparate impacts. The legislation would require independent evaluations of algorithms before they are deployed, along with ongoing assessments to identify and mitigate bias over time. It would also mandate transparency, including notifying individuals when an AI system is used in decision-making and providing a pathway to appeal automated decisions to a human reviewer. Enforcement authority would rest with the Federal Trade Commission (FTC), state attorneys general, and individuals who can show harm.

Civil rights and consumer advocacy groups have welcomed the bill, arguing that existing anti-discrimination laws have not kept pace with the growing use of automated systems. Some supporters argue that federal standards are critical amid congressional discussions to limit or preempt state-level AI regulation.

However, the proposal has drawn criticism from industry groups, some legal scholars, and technology policy advocates. Critics argue that the bill’s reliance on disparate impact standards could expose companies to broad legal liability even when no intentional discrimination is present. Others warn that mandatory audits, documentation, and appeal mechanisms could be costly and difficult to implement, particularly for smaller firms, potentially slowing innovation or discouraging the use of AI in beneficial applications.

Some opponents also contend that existing civil rights, consumer protection, and employment laws already provide tools to address discriminatory outcomes, and that new AI-specific rules could create overlapping or unclear obligations. Questions have also been raised about how the bill would be enforced in practice and whether federal requirements could conflict with sector-specific regulations.

The bill’s prospects in Congress remain uncertain, and it is unclear whether it will gain bipartisan support. Its reintroduction nonetheless reflects continued debate over how far federal lawmakers should go in regulating AI systems and addressing their social and economic impacts.