Proptech and Housing Rights: Promise and Peril

A recent piece in The Regulatory Review argues that property technology, the software now sitting between people and almost every housing decision, can undermine housing fairness and human rights (The Regulatory Review, October 15, 2026). I want to take that argument seriously and push on it a little. The interesting question, in my view, is not whether proptech is good or bad. It is what happens to a right when the people who used to exercise judgment about it are replaced by a score.

What is proptech, and why does it matter for housing rights?

Proptech is the broad label for software that runs the housing market: tenant screening, rent-setting tools, online listing and advertising platforms, automated valuation models, digital leasing, and the systems landlords use to manage eviction filings. A decade ago most of these decisions were made by a leasing agent, a loan officer, or a property manager, which meant they were made by someone who could be asked a question, corrected, or sued.

Housing is different from most consumer markets because it is tied to rights that are already written down. In the United States the Fair Housing Act, 42 U.S.C. § 3601 and following, prohibits discrimination in the sale, rental, and financing of housing based on race, color, religion, sex, familial status, national origin, and disability. Internationally, the right to adequate housing sits in Article 11 of the International Covenant on Economic, Social and Cultural Rights. Proptech does not repeal any of this. What it does is move the decision somewhere the law has a harder time reaching.

That is the heart of the promise and the peril. The same tool that can remove a biased human gatekeeper can also hide a biased pattern inside a model nobody can interview.

What is the promise?

I do not want to wave away the upside, because it is real. A tenant who has been rejected by three landlords for reasons nobody would explain has a legitimate complaint about the old system. Software can, in principle, do several things better.

• Consistency. A rule applied the same way to every applicant is easier to audit than the gut feeling of a hundred different leasing agents.
• Access. Online applications, digital documents, and virtual tours lower the cost of searching for housing, especially for people who cannot take a day off to visit properties.
• Alternative data. Rent payment history reported to credit bureaus can help renters with thin credit files show they are reliable.
• Speed. Faster approvals and faster repairs matter a great deal to families who are one missed deadline away from losing a place.

A consistent rule is only as fair as the rule. That is where the trouble begins.

Where does the peril show up?

The Regulatory Review piece frames proptech as a threat to housing fairness and human rights. When I look at the record of litigation and enforcement over the last several years, the problems cluster in four places.

Advertising and who gets to see a listing

In March 2019 the Department of Housing and Urban Development charged Facebook with violating the Fair Housing Act, alleging its advertising tools let advertisers exclude people by protected characteristics and that its delivery system itself steered housing ads. In June 2022 the Justice Department announced a settlement with Meta in United States v. Meta Platforms, Inc. (S.D.N.Y.), under which Meta agreed to stop using its housing ad targeting tool and to build a new system, called the Variance Reduction System, to reduce delivery disparities.

The lesson here is that discrimination does not require anyone to type a discriminatory instruction. An optimization system trained to maximize clicks can learn that certain people click on certain listings and quietly show everyone else less.

Tenant screening and the opaque score

Tenant screening reports now combine credit data, eviction records, criminal records, and proprietary risk scores. Those reports are regulated in part by the Fair Credit Reporting Act, 15 U.S.C. § 1681 and following, which gives applicants the right to be told when a report was the basis for an adverse decision and to dispute inaccurate information.

In Louis v. SafeRent Solutions, filed in the District of Massachusetts, Black and Hispanic housing voucher holders challenged a scoring system they said gave too little weight to the reliability of voucher-based rent and too much to credit history. The case settled in 2024 with a payment and an agreement to change how the product treats voucher holders. I find the case instructive because the score looked neutral. It was a number. Nobody had to say anything about race for the number to land harder on some applicants than others.

Algorithmic rent-setting

The pricing side raises a different question. In August 2024 the Justice Department, joined by several state attorneys general, sued RealPage in the Middle District of North Carolina, alleging that its revenue management software let competing landlords effectively coordinate rents by feeding sensitive, nonpublic data into a common pricing tool. That is an antitrust claim, not a fair housing claim, but the housing rights implications are obvious. If a shared algorithm nudges rents upward across a whole market, the people who bear the cost are the tenants with the fewest alternatives.

Whether the theory ultimately succeeds is for the courts and the parties to resolve. What I find worth noticing is that regulators decided to treat housing pricing software as something that could be a market-wide problem and not merely a vendor's private business.

Valuation and lending

Automated valuation models and underwriting tools carry the longest history of all, because they inherit a century of data shaped by redlining and unequal investment. A model trained on past prices and past approvals will faithfully reproduce the past. I have written elsewhere about how AI often exposes patterns that were always there, and housing may be the clearest example. The software did not invent the pattern. It made the pattern run faster and look more objective.

How do the legal standards compare?

The legal tools available to challenge proptech differ in what they require a plaintiff to prove, and that difference shapes everything that follows.

The row that matters most to me is disparate impact. In Texas Department of Housing and Community Affairs v. Inclusive Communities Project, Inc., 576 U.S. 519 (2015), the Supreme Court held that disparate impact claims are cognizable under the Fair Housing Act. That decision is what lets a plaintiff challenge a neutral-looking algorithm without proving that anyone intended harm. Without it, most proptech discrimination would be nearly unreachable.

Is the legal ground shifting?

Yes, and not in one direction. On April 23, 2025, an executive order titled "Restoring Equality of Opportunity and Meritocracy" directed federal agencies to deprioritize enforcement of disparate impact liability. For an area of law where the algorithm's effect is often the only visible evidence, that is a significant change in federal posture. Private lawsuits and state enforcement remain available, and the Supreme Court's 2015 holding on the statute has not changed, but the weight of federal enforcement matters because individuals rarely have the data or resources to prove a pattern across thousands of applicants.

I covered the broader federal reshuffling in this week-in-review piece, and the housing context fits the same story: when federal enforcement recedes, states, cities, private plaintiffs, and the vendors themselves become the main source of discipline. That tends to produce a patchwork, and patchworks are hardest on the people with the least capacity to navigate them.

If you want a wider frame for how these questions travel across jurisdictions, my piece on technology policy is a good place to start.

What does this mean for landlords, lenders, and platforms?

The reader of a news response usually wants to know what to do with it, so here is my analysis of what the current record implies for different players. These are my judgments, not legal advice.

For landlords and property managers. Buying a screening or pricing tool does not transfer liability to the vendor. The Fair Housing Act applies to the housing provider, and the cases above name both vendors and users. Asking a vendor what inputs go into a score, whether the model has been tested for disparate outcomes, and how an applicant can contest a result seems to me the minimum. If the vendor cannot answer, that is information.

For lenders and valuation firms. Models trained on historical data need ongoing testing against the populations they will actually affect. A model that was accurate in aggregate can still be wrong in a systematic way for a specific neighborhood.

For platforms. The Meta settlement shows that a company can be held responsible for effects produced by its delivery system, not only for the options it offers advertisers. Optimization is a design choice.

For tenants and advocates. The Fair Credit Reporting Act's dispute and notice rights are underused. An adverse action notice names the reporting agency, and from there an applicant can request the file. It is a modest tool, but it is concrete.

For policymakers. The harder question is where to put the burden of proof. Right now it sits almost entirely on the person who was rejected, who cannot see the model and often does not know a model was involved.

Why the human rights framing matters

The Regulatory Review piece describes the problem in human rights terms, and I think that framing does real work. A consumer protection lens asks whether a product was accurate and whether the disclosure was adequate. A rights lens asks whether a person was treated as someone with a claim on a home, or as a risk profile.

Those are different questions. A screening algorithm can be accurate, disclosed, and still deny a family a place to live for reasons that have nothing to do with whether they would pay rent or care for the property. Accuracy about the wrong thing is a very efficient way to do harm.

There is also a subtler cost. When a decision is delegated to software, nobody feels responsible for it. The leasing agent says the system flagged it. The vendor says the landlord set the threshold. The landlord says the vendor's score is industry standard. I wrote about this dynamic in Thinking Without Proxies, and housing is where the stakes of a proxy are clearest. A score stands in for a judgment, and the judgment quietly disappears.

What should we be asking?

I do not think the answer is to ban software from housing. The old system had its own cruelty, and some of it was hidden in exactly the discretion that people now miss. What I have come to think is that a few questions should travel with every proptech tool.

• Who can see how a decision was made, and who can challenge it?
• What was the model trained on, and does that history include discrimination we already know about?
• Who pays when it is wrong, the person denied or the company that deployed it?
• Can a human with real authority override the result?

If a tool cannot survive those questions, it probably should not be deciding where someone lives. And if we find that the only way to make a system efficient is to make it unanswerable, we have to ask what we were optimizing for in the first place.

Housing is the one market where the product is also the place a person sleeps, raises children, and builds a life. Software that treats it like any other catalog item has already misunderstood the job.

Jared Clark, Founder of Prepare for AI

Last updated: 2026-10-24

Frequently Asked Questions

What is proptech?

Proptech, or property technology, is software used in the housing and real estate market. It includes tenant screening, rent-setting tools, online listing and ad platforms, automated valuation models, and digital leasing and property management systems.

How can proptech violate the Fair Housing Act?

The Fair Housing Act, 42 U.S.C. § 3601 and following, bars housing discrimination based on protected characteristics. Proptech can violate it through ad targeting or delivery that excludes protected groups, screening scores that disproportionately reject them, or models trained on discriminatory historical data. Under Texas Department of Housing and Community Affairs v. Inclusive Communities Project (2015), a plaintiff can bring a disparate impact claim without proving intent.

Are landlords liable for discriminatory screening software they did not build?

Generally, the Fair Housing Act applies to housing providers regardless of whether a vendor built the tool. Landlords who rely on a screening product can face claims, and vendors have also been named as defendants, as in Louis v. SafeRent Solutions. Asking vendors about testing and dispute processes is a practical safeguard.

What is the RealPage case about?

In August 2024 the U.S. Department of Justice and several state attorneys general sued RealPage in the Middle District of North Carolina. They alleged its revenue management software let competing landlords coordinate rents through a shared pricing tool using nonpublic data. It is an antitrust case, but it bears directly on housing affordability.

What can a tenant do after an automated screening denial?

Under the Fair Credit Reporting Act, 15 U.S.C. § 1681 and following, an applicant denied because of a screening report is entitled to notice naming the reporting agency, and can request their file and dispute inaccurate information. This does not guarantee a fair outcome, but it is the most concrete route available.