The SafeRent litigation raised a specific question for housing providers: what happens when a screening score influences a rental decision but does not adequately reflect the applicant’s circumstances?
This article was first published on 6 December 2024, shortly after the settlement received final approval. The case concerns algorithmic tenant screening. It should not be treated as a finding that every AI product is discriminatory, or as proof that a vendor’s stated ethical principles ensure fair results.
What the case resolved
In Louis et al. v. SafeRent Solutions, applicants challenged the use of SafeRent Scores in Massachusetts, including their effect on applicants using housing vouchers and Black and Hispanic voucher holders. The litigation alleged violations of the Fair Housing Act and Massachusetts law. SafeRent denied wrongdoing.
The court’s account of the allegations explains the proposed mechanism. Plaintiffs said the score used credit history, including debts unrelated to tenancy, without accounting for the portion of rent covered by a housing voucher. They alleged that racial disparities in credit histories translated into lower scores and housing denials for Black and Hispanic voucher holders. Housing providers could set approval thresholds but could not change the underlying algorithm or see its factor weights. The concern was that a low score could trigger rejection without considering the rent-payment circumstances the plaintiffs said mattered. This describes their allegations, not a final finding of liability.
The agreement provided total settlement consideration of up to $2.275 million, including compensation, administration and specified awards and legal costs. That amount is not solely money distributed to applicants. The settlement agreement defines the classes and terms.
The administrator lists 20 November 2024 as the date final approval was granted, following a hearing on 18 November. It reports check payments mailed on 30 June 2025 and electronic-payment emails sent on 1 July 2025. For eligible members who chose two installments, the second payments were issued on or before 17 February 2026. The official settlement website provides the status and documents; the original claim deadline has passed.
Settlement approval resolved the claims against SafeRent under the agreement. It should not be described as a trial finding that SafeRent committed the alleged discrimination.
The practice changes had defined limits
The five-year practice changes restrict scores and score-based recommendations in SafeRent’s affordable model. For market and no-credit models, they apply unless the customer certifies that the applicant is not using a publicly funded housing voucher. The agreement permits underlying information and defines exceptions for validated scores; it does not permanently ban all scoring.
Housing providers should review their model, current vendor instructions and section 3.5 of the agreement with qualified advisers.
Review the decision process behind a score
The practical lesson is to understand what your screening process actually does. A vendor supplies information or a recommendation, but your company still needs an accountable process for applying approved criteria, reviewing issues and communicating the result.
Begin with a recent, appropriately selected sample of files.
Check:
- Which product and configuration were used.
- What information influenced the decision.
- Who reviewed it.
- What the applicant was told.
A simple “failed screening” note may not explain whether the issue was inaccurate data, an inappropriate criterion or a decision that was applied incorrectly.
Ask the vendor questions that produce usable answers:
| Area | Question to resolve |
|---|---|
| Inputs | What data influences the report or recommendation, and how is it verified? |
| Intended use | Which applicants and housing circumstances was the product designed and evaluated for? |
| Configuration | Which criteria come from the vendor and which are set by the housing provider? |
| Explanation | Can the team identify the information behind an unfavorable recommendation? |
| Correction | How can a disputed record be investigated and corrected? |
| Change management | How will you be informed about material model, data or product changes? |
| Oversight | What evidence can qualified reviewers examine about performance and potential unfair effects? |
A claim that a system is transparent or unbiased is not an answer to those questions. Transparency can help investigation, but it does not establish that a criterion is appropriate or that the resulting decisions comply with the law.
Make applicant communications part of the workflow
The FTC explains that tenant-screening reports are consumer reports. When a housing provider takes an adverse action based partly or wholly on a consumer report, notice obligations apply. Adverse action can include a denial or less favorable terms, such as requiring a larger deposit. The FTC’s landlord guidance sets out the requirements, including additional obligations when a credit score is used.
Build those requirements into a reviewed procedure. Identify who makes the decision, who prepares the required communication and where completion is recorded. A generic rejection email is not a substitute for establishing which obligations apply.
An applicant who identifies a possible error needs a clear route to the appropriate person and reporting company. Staff should know how to preserve the relevant record, explain the established process and avoid making promises about an outcome they cannot authorize. Keep records secure and limit access to people who need them.
Give human review a real function
In an illustrative case, a report contains information that the applicant says belongs to another person. A meaningful reviewer examines the dispute through the approved process and obtains the relevant evidence. Repeating the automated result without investigating would add little protection.
Set screening policies with advisers familiar with the jurisdictions and housing programs you serve. Apply approved criteria consistently, and establish a documented route for questions, corrections and matters requiring individualized review. Those responsibilities remain important regardless of whether a product is marketed as AI, an algorithm or a conventional screening service.
Our AI use-case guide separates administrative assistance from consequential decisions. The tenant-management guide connects communication and record ownership across the resident relationship.
Where approved screening procedures are being applied inconsistently because responsibilities are unclear, BetterTeam Consulting can help clarify the team’s operating roles. Legal criteria and case-specific advice belong with qualified counsel.
