Why It Matters
Federal anti-discrimination law leaves significant questions about algorithm-based decisions unresolved, creating uncertainty for employers, creditors, housing providers, and others who use automated tools, according to a new Congressional Research Service (CRS) report. The report says the law "is unsettled on how to evaluate cases where no human consciously chooses to discriminate, and yet an automated system classifies people on the basis of protected characteristics."
Executive Order 14281, which directs agencies to deprioritize disparate-impact enforcement, may constrain how federal agencies address algorithmic discrimination, the report says. That combination leaves Congress with a choice: clarify how existing civil rights protections apply to automated decisions or leave the framework to courts and an administration moving to narrow disparate-impact enforcement.
The Big Picture
The report identifies unresolved questions under two major anti-discrimination theories. For disparate treatment, it examines whether discrimination through proxy variables requires proof of human intent and warns that plaintiffs "may never be able to prove disparate treatment" when an AI model's decisions cannot be explained. For disparate impact, questions include whether an algorithm constitutes a "specific practice," what justification defendants must provide, and how plaintiffs can identify less discriminatory alternatives.
The Trump administration has moved on several fronts. A Justice Department Office of Legal Counsel opinion issued in June concluded that the Equal Employment Opportunity Commission's (EEOC) interpretation of Title VII's disparate-impact provisions is unconstitutional. The Department of Housing and Urban Development (HUD) has proposed rescinding its Fair Housing Act disparate-impact regulations and related Title VI disparate-impact regulations. Both remain proposals. The EEOC has also proposed eliminating certain reporting requirements, citing constitutional concerns.
The report says Congress could amend existing antidiscrimination statutes or establish standalone requirements for transparency, bias audits or human participation in algorithmic decisions. It also cautions that some legislative approaches could raise constitutional questions under the Takings Clause, First Amendment, and Equal Protection Clause.
Five bills introduced in the 119th Congress remain at the committee-referral stage: the Algorithmic Accountability Act of 2025 (H.R. 5511), sponsored by Rep. Yvette Clarke (D-NY); S. 2164, sponsored by Sen. Ron Wyden (D-OR); the Artificial Intelligence Civil Rights Act of 2025 (H.R. 6356), also sponsored by Clarke; S. 3308, sponsored by Sen. Edward Markey (D-MA); and the Right to Override Act (S. 2997), also sponsored by Markey.
The Bottom Line
Congress could clarify how civil rights protections apply to algorithmic decisions by amending existing laws or adding new transparency, auditing, and human-oversight requirements. Meanwhile, unresolved questions about discriminatory intent, proxy discrimination, and disparate impact are becoming more consequential as the Trump administration seeks to narrow federal enforcement. HUD's proposals would remove regulatory disparate-impact protections in housing and federally assisted programs, while the EEOC's proposed changes would reduce certain employer reporting obligations.
Access the Legis1 platform for comprehensive political news, data, and insights
Spot something wrong? Report an issue with this article
