What to Do in Algorithmic Discrimination Cases

You didn't get the job. Or the loan. Or the apartment. And somewhere in the background, a piece of software made the call — not a person.

This is happening to more Americans every year. Resume-screening tools, credit-scoring models, tenant-screening software, and insurance algorithms now shape decisions that used to sit entirely with a human being. The problem: these systems can absorb and repeat the same biases people have, just faster and at a much bigger scale. And unlike a biased manager, an algorithm doesn't leave an obvious paper trail of "why."

If you suspect this happened to you, here's what it actually means, and what you can do about it, step by step.

What Counts as Algorithmic Discrimination

Algorithmic discrimination happens when an automated system — often powered by AI or machine learning — produces an outcome that unfairly disadvantages people based on a protected trait: race, sex, age, disability, national origin, religion, or similar categories.

It doesn't require anyone to have programmed the bias on purpose. Most cases involve disparate impact — a facially neutral tool that still produces lopsided results. A resume screener trained on ten years of a company's hiring data might learn to favor traits common among past hires, quietly filtering out equally qualified applicants who don't fit that pattern. Nobody intended that. It happened anyway, and the law generally treats the outcome as the problem, not the intent behind it.

You'll typically encounter this in four areas:

• Employment — resume scanners, video-interview scoring, chatbot screening, scheduling and promotion tools

• Credit and lending — automated underwriting and credit-scoring models

• Housing — tenant-screening algorithms and mortgage approval systems

• Insurance — pricing and claims models that vary rates by proxy factors

The Law You Can Actually Use Today

Here's the part that surprises most people: there's no single federal "AI discrimination law" in the US. Instead, existing civil rights laws already cover algorithmic decisions, because they're written around the outcome, not the method.

Federal law that already applies:

• Title VII of the Civil Rights Act — bars discrimination in employment based on race, color, religion, sex, or national origin, whether a human or an algorithm made the call

• The Americans with Disabilities Act (ADA) — requires that AI hiring tools not screen out qualified people with disabilities, and that accommodations be offered

• The Age Discrimination in Employment Act (ADEA) — covers workers 40 and older

• The Equal Credit Opportunity Act (ECOA) and Fair Housing Act — extend to automated lending and housing decisions

The Equal Employment Opportunity Commission has been explicit that algorithmic hiring tools are subject to the same federal anti-discrimination statutes as human decision-makers, and relying on the software as an excuse is not a valid legal defense. Worth noting: detailed EEOC technical guidance on how these laws apply to AI was taken down from the agency's website in early 2025 after a shift in federal AI policy. But that's guidance, not law — the underlying statutes are unchanged, and the removal of that guidance doesn't change the underlying requirements of Title VII, the ADA, or state-level equivalents.

A growing patchwork of state laws. Since federal rulemaking has slowed, states have moved in. As of 2026:

• Colorado's AI Act requires companies that build or deploy "high-risk" AI systems (used in employment, lending, housing, and more) to run risk assessments, notify people when AI is used in a decision about them, and report discovered discrimination to the state attorney general — though this law is currently being challenged in federal court by an AI company, with the Justice Department intervening in support of the challenge, so its future is genuinely unsettled.

• Illinois amended its Human Rights Act so employers may not use AI in a way that has the effect of discriminating against employees or applicants at any stage, from recruitment through termination.

• New York City's Local Law 144 requires bias audits for automated employment-decision tools used on NYC candidates.

• Texas, California, and several other states have layered on their own disclosure or anti-bias rules for AI in hiring, lending, and beyond.

Because this area is moving fast and some laws are tied up in litigation, treat any specific state law as something to verify against current news before you rely on it — what's on the books today may look different in a few months.

Signs You May Have Experienced It

You often won't get a clear explanation, but a few patterns are worth noticing:

• You were rejected instantly or within minutes of applying, with no human contact

• You were asked to complete a video interview, game-based assessment, or personality test scored automatically

• A denial letter mentions a "credit model," "automated system," or "algorithm" rather than a specific reason

• You know your qualifications were strong relative to the role, yet you were screened out at the very first stage

• Others in your demographic group have reported similar rejections from the same company or platform

None of these alone proves discrimination. But together, they're a reason to dig further.

What to Do, Step by Step

1. Document everything immediately

Save the job posting, application, any denial or rejection notice, emails, and screenshots. If you were scored or assessed by software, note the platform name if it's disclosed (some job postings or emails mention it). Write down dates and any conversations. This record matters far more than memory once weeks have passed.

2. Request an explanation

You can ask the company directly what tool was used and why you were rejected. Under some state and local laws (like NYC's), employers using automated hiring tools are required to disclose that fact. For credit denials, the Equal Credit Opportunity Act already requires lenders to give you specific reasons — ask for it in writing (an "adverse action notice") if you didn't receive one automatically.

3. Check whether a bias audit or disclosure was legally required

If you're in a jurisdiction with an AI hiring law (NYC, Illinois, Colorado, California, Texas), the employer may have had a legal obligation to test the tool for bias or notify you it was in use. If they didn't, that's itself a violation you can raise.

4. File a complaint with the right agency

• Employment: File with the EEOC (1-800-669-4000, or the EEOC Public Portal) within the applicable deadline — typically 180 or 300 days from the discriminatory act depending on your state, so don't sit on this.

• Credit/lending: File with the Consumer Financial Protection Bureau (CFPB) or the Federal Trade Commission (FTC).

• Housing: File with the Department of Housing and Urban Development (HUD) or your state fair housing agency.

• State-level claims: Many states have their own civil rights or human rights commissions that accept complaints faster than federal agencies and may have their own AI-specific rules.

5. Consider an employment or civil rights attorney

Algorithmic discrimination cases are technically complex — proving disparate impact often involves statistical analysis of who got screened out. A lawyer experienced in employment or consumer law can request the data and expert analysis needed to build the case, and can tell you whether you might be able to join an existing class action. Several major cases (Workday's AI screening tool is one widely reported example) have already moved forward as class actions, which means you may not be alone even if it feels that way.

6. Ask about human review

Where the law requires it (Colorado's law is one example), you may be entitled to ask for a human to review an automated decision, and to appeal it. Even where it's not legally required, requesting human review in writing creates a record that you tried — and companies sometimes reverse automated decisions once a person actually looks.

What You're Not Required to Prove

You don't need to prove the company intended to discriminate. Under disparate impact analysis, showing that a neutral-looking tool produced a lopsided outcome for a protected group can be enough to shift the burden to the employer or lender to justify the tool as job-related, necessary, and lacking a less discriminatory alternative. This is one of the more employee-friendly aspects of existing civil rights law, and it applies whether a person or an algorithm made the call.

The Bottom Line

Algorithmic discrimination sits at an unusual point in US law right now: the core protections — Title VII, the ADA, ECOA, the Fair Housing Act — already apply, but the specific federal guidance on how they apply to AI has been pulled back, while states are rushing in with their own rules, some of which are being challenged in court. That mix can feel confusing, but it doesn't leave you without options. Document what happened, ask for the reasons behind the decision, and file with the right agency. The law hasn't caught up perfectly to the technology, but it hasn't been left behind either.