AI, Algorithms and Fair Housing: What Housing Providers Need to Consider

Artificial intelligence can help property managers accomplish tasks that once required hours of manual work.

Applications can be reviewed more quickly.

Documents can be analyzed automatically.

Potential fraud can be flagged.

Screening information can be organized and compared against predefined criteria.

For busy leasing teams, that sounds promising.

But there is another side of the conversation.

When algorithms influence who receives housing, property managers need to think carefully about Fair Housing, data accuracy, transparency, and how automated decisions are made.

The technology may be new.

The responsibility to avoid unlawful housing discrimination is not.

Fair Housing Rules Don't Disappear When Technology Is Involved

The Fair Housing Act prohibits discrimination in housing because of race or color, religion, sex, national origin, familial status, and disability.

Using software, an algorithm, or an outside screening provider does not make those underlying legal obligations irrelevant.

In 2024, HUD specifically addressed algorithmic tools and tenant screening in agency guidance, illustrating regulators' concern about the intersection of automated systems and housing decisions. Some federal guidance documents have subsequently been reviewed or withdrawn as agency policy changed, which is also why housing providers should rely on current law and qualified legal counsel rather than treating older guidance as permanent compliance rules.

The practical lesson remains useful:

Understand how technology affects your screening decisions.

How Could an Algorithm Affect Fair Housing?

Automated systems evaluate information according to programmed rules, models, or patterns.

Depending on the system, inputs might include:

• Credit information
• Rental history
• Eviction records
• Criminal records
• Income information
• Address history
• Identity information
• Other applicant data

The system may then generate:

• A score
• A risk category
• A recommendation
• A pass/fail result
• A suggested condition for approval

The potential issue is not simply whether the computer knows an applicant's race, religion, disability, or another protected characteristic.

The broader question is:

What information is being used, and what effect does the screening process have?

An Algorithm Can Feel Objective Without Being Easy to Understand

People tend to trust numbers.

If software assigns an applicant a score of 742, for example, the number can feel scientific and precise.

But what does 742 actually mean?

How was it calculated?

Which factors mattered most?

Would changing one input dramatically change the outcome?

Has the system been evaluated for accuracy?

Could incomplete or mismatched information affect the score?

Property managers do not necessarily need to become data scientists.

But they should understand enough about the tools they use to know what role those tools play in housing decisions.

Watch Out for the “Black Box” Problem

A black-box system produces a result without giving the user a meaningful understanding of how the result was reached.

Imagine an applicant is denied because an automated platform labels the application “high risk.”

The leasing team does not know why.

The property manager does not know why.

The applicant certainly does not know why.

That creates practical problems even before considering legal issues.

How does your team:

• Confirm the information is accurate?
• Determine whether your rental criteria were applied correctly?
• Respond when the applicant disputes a record?
• Identify whether a data source is wrong?
• Explain what contributed to an adverse decision?

Transparency matters.

Bad Data Can Become an Automated Problem

Algorithms rely on data.

Tenant screening data can sometimes contain errors, including mismatched identities, outdated information, or records that applicants contend are inaccurate. The CFPB has documented consumer complaints related to inaccurate tenant background checks, and the FTC provides information about renters' FCRA rights concerning these reports.

Consider what happens if incorrect data enters an automated system.

The software may analyze it perfectly.

But the result can still be wrong.

Examples might include:

• An eviction record belonging to someone with a similar name
• A criminal case that was dismissed
• An outdated debt balance
• An incorrect address
• A record connected to the wrong individual

Automation does not eliminate the need for accurate data.

In some cases, it makes accurate data even more important because decisions can happen more quickly and at greater scale.

Be Careful With Proxy Information

Algorithms can identify relationships between pieces of information that people might never notice.

That is part of what makes advanced analytics powerful.

But housing providers should understand which factors are being considered and whether those factors are appropriate for housing decisions.

A variable does not necessarily need to explicitly identify a protected characteristic to deserve scrutiny.

For example, organizations should think carefully about the relevance of data such as:

• Geographic information
• Highly granular neighborhood information
• Certain behavioral indicators
• Nontraditional consumer data
• Other variables unrelated to established rental qualifications

The core question should remain:

Why is this information relevant to our legitimate rental criteria?

If that question is difficult to answer, it may be worth discussing with the provider and legal counsel.

Don't Assume the Vendor Has Solved Everything

Property managers often rely on technology vendors because those vendors specialize in screening.

That can be helpful.

But outsourcing the technology does not mean a housing provider should stop asking questions.

When evaluating an automated screening provider, consider asking:

• Which applicant information does the system evaluate?
• Where does the information come from?
• How are records matched?
• Does the system produce a recommendation or make an automatic decision?
• Can we customize the criteria?
• How are errors corrected?
• What happens when an applicant disputes a record?
• Can we see the underlying information?
• What testing or validation has been performed?
• How does the provider address applicable consumer-reporting requirements?

You may not receive access to every detail of a proprietary model.

But you should understand what you are relying on.

FCRA Obligations Matter Too

Fair Housing is not the only issue surrounding automated tenant screening.

When a tenant screening service provides a consumer report for housing eligibility, the Fair Credit Reporting Act may apply. The FTC identifies tenant screening services as among the entities covered by the FCRA when they provide qualifying consumer reports.

That brings issues such as:

• Permissible purpose
• Accuracy
• Consumer disputes
• Adverse action

into the conversation as well.

Technology may change how information is processed, but it does not erase consumer-reporting responsibilities.

Written Rental Criteria Become Even More Important

One of the best protections against inconsistent screening is having clearly established rental criteria.

Those criteria can help determine whether your technology is doing what you expect it to do.

Ask:

• What factors do we actually consider?
• Are those factors reflected in our written policy?
• Is the automated system using additional information we did not intend to consider?
• Does the technology apply the same standards to similarly situated applicants?
• Can our employees override results?
• If so, when and how are overrides documented?

Without clear rental criteria, it is difficult to evaluate whether an algorithm is helping or quietly changing your process.

Human Overrides Need Rules Too

Some organizations respond to concerns about automated decisions by allowing employees to override them.

That sounds like an obvious safeguard.

But uncontrolled overrides can create another form of inconsistency.

Imagine:

Applicant A receives an unfavorable automated recommendation.

A leasing agent likes Applicant A and overrides the system.

Applicant B receives the same result.

A different leasing agent does not override it.

Now two similarly situated applicants may have been treated differently.

Human review can be valuable.

But it should operate according to established procedures rather than personal preference.

Consider Regularly Auditing the Process

Housing providers periodically review leases, pricing policies, security procedures, and vendor contracts.

Automated screening tools deserve the same attention.

An audit might ask:

• Are we still using the same screening criteria?
• Has the provider changed its model or data sources?
• Are applicants frequently disputing certain records?
• Are staff members overriding results?
• Are adverse action notices being handled correctly?
• Are there significant differences in outcomes that deserve review?
• Are state or local requirements changing?

You do not have to wait for a complaint to examine how the process is working.

Local Laws Can Be More Restrictive

Automated screening is only one layer of an increasingly complicated regulatory environment.

State and local jurisdictions may have additional restrictions affecting matters such as:

• Criminal history
• Eviction history
• Credit information
• Income
• Source of income
• Application procedures
• Applicant disclosures

The rules affecting one property may differ from those affecting another property in a different city or state.

That makes legal review particularly important for housing providers operating across multiple jurisdictions.

Technology Should Support Fair and Consistent Decisions

The goal does not need to be avoiding automation.

Automated tools can provide meaningful operational benefits.

They can help standardize workflows, identify inconsistencies, reduce administrative workload, and potentially improve the application experience.

The question is how they are used.

A strong approach keeps several principles in focus:

• Use relevant information
• Understand your screening criteria
• Verify data quality
• Maintain appropriate human oversight
• Provide a process for disputes
• Apply standards consistently
• Periodically evaluate vendors and technology

The Bottom Line

AI and algorithms will likely continue to play a role in rental housing.

Property managers do not need to fear that technology.

But they should understand it.

The fact that a decision came from software does not automatically make it accurate, neutral, or appropriate.

Key Takeaway

When evaluating an automated tenant screening tool, ask:

“Can we explain what information this system uses, why we use it, and how it affects our rental decisions?”

If your organization cannot answer those questions, it may be time to learn more about the technology behind the screening process.

AI can be a valuable tool.

The strongest screening programs will use it alongside clear criteria, accurate information, appropriate oversight, and a continued commitment to consistent and lawful housing decisions.

This article is provided for general informational purposes and is not intended as legal advice. Federal agency guidance and enforcement policies can change, and state and local laws may impose additional requirements. Housing providers should consult qualified counsel regarding the use of automated technology, consumer reports, and screening criteria.