Compliance

Fair Housing and AI Screening: Keeping Applicant Screening Defensible

HUD says the Fair Housing Act applies to tenant screening, algorithms included. Here is what the 2024 guidance asks of you, and how to keep screening defensible.

By The Access · June 21, 2026 · 8 min read

If you screen rental applicants, the Fair Housing Act applies to how you do it, and HUD has made clear that does not change when an algorithm makes part of the call. In 2024, HUD's Office of Fair Housing and Equal Opportunity issued guidance spelling out how the law applies to tenant screening, including the growing use of third-party screening companies and AI. The message was simple: you are responsible for the outcome, whoever or whatever produced it.

The two ways screening goes wrong under the Act

The Fair Housing Act prohibits two distinct things. The first is intentional discrimination, treating an applicant worse because of a protected characteristic. The second is harder to spot: a policy that looks neutral on its face but produces an unjustified discriminatory effect on a protected group. That second one, disparate impact, is where most screening problems live, because a blanket rule can disadvantage a protected class without anyone intending it to.

The three-step test HUD applies

When a screening practice is challenged for disparate impact, the analysis runs in three steps. First, the applicant has to show the practice causes a disparate impact on a protected group, or reinforces segregation. Second, the burden shifts to you to show the practice is necessary to achieve a substantial, legitimate, nondiscriminatory interest. HUD is explicit that only criteria well-tailored to predict behavior actually relevant to tenancy can clear this bar. Third, the burden shifts back, and the applicant can still win by showing a less discriminatory alternative would serve the same interest. If your screen is overbroad, meaning it rejects applicants who would have been fine, a more targeted screen is the less discriminatory alternative, and you lose.

The three pieces of a screening report, and where each goes wrong

What HUD asks you to do

The guidance reads like a checklist for a defensible process. Use only screening criteria that are actually relevant to tenancy. Publish your screening policy in advance so applicants know the standards. Keep human discretion over third-party results instead of treating a pass or fail grade as the decision. Give applicants a real chance to contest a negative determination and correct bad data. Make sure any model you rely on is accurate and non-discriminatory, and avoid sloppy data practices like name-only or wildcard matching that attribute records to the wrong person. When you deny, provide the records you relied on and the specific reasons. And account for context, the age of an offense, an explanation, a circumstance that has changed.

The part people want to skip

You stay liable even when a third party does the screening. Hiring a vendor, or pointing at an algorithm, does not move the responsibility off you. That is the whole reason a defensible process matters: it is not just fairer, it is the thing that protects you when a decision gets questioned.

Defensible screening comes down to two things working together, relevant criteria and a human in the loop with the context to apply them. The Access keeps your stated standards and your own documents in front of the decision, so screening follows the policy you can defend, not an opaque score nobody can explain.

This is an operations summary, not legal advice. The rules, notices, thresholds, and dates described here are set by HUD and other agencies and they change. Confirm the current requirements against the official notices and with your compliance team or counsel before you rely on them.

Screening you can stand behind

Keep the policy in front of the decision

The Access works from your stated standards and your own documents, so screening stays consistent, explainable, and yours.

Keep reading

Related