The Algorithm Said No: How Tenant Screening Algorithms Influence Housing Decisions
An overview of data sources, legal context, and practical considerations for renters and housing providers
Introduction: When a Score Becomes a Barrier
Many rental applications are evaluated using automated tenant screening tools. In some cases, applicants are denied housing based on a score or recommendation without receiving a detailed explanation of what factors contributed to the decision. As these systems become more common, regulators, courts, and consumer advocates have raised questions about transparency, data accuracy, and how consumer information is used in housing decisions.
Tenant screening has become a standard part of the rental application process in the United States. Many landlords rely on third-party screening companies that use automated systems to assess applications and summarize risk using scores or recommendations.
This article explains how tenant screening algorithms work, the types of data they rely on, recent legal and regulatory developments involving these systems, and the practical implications for renters.
CASE STUDY: THE SAFERENT LAWSUIT
Mary Louis applied for an apartment after paying rent consistently for more than a decade and maintaining stable income. As part of the application process, the landlord used SafeRent, a tenant screening company, to evaluate her application.
SafeRent generated a low score, and the landlord denied the application. Louis was not provided with a detailed explanation of how the score was calculated.
In November 2024, a federal court approved a 2.275 million dollar settlement in Louis v. SafeRent Solutions, LLC. The case examined whether SafeRent’s scoring practices complied with the Fair Housing Act, particularly in their treatment of applicants using housing vouchers.
The settlement did not require an admission of wrongdoing, but it did require changes to SafeRent’s screening practices nationwide.
The Tenant Screening Industry
Tenant screening is a large and growing industry, estimated at approximately 1.3 billion dollars annually. Major providers include:
- SafeRent (formerly CoreLogic)
- TransUnion
- RentGrow
- Other regional and specialized firms
These companies typically aggregate data from multiple sources and apply automated models to summarize information for landlords.
Common data inputs include:
- Credit reports and credit scores
- Eviction court records
- Criminal background data
- Income and employment verification
- Rental payment history
- Public records such as bankruptcies and liens
The output is often a numerical score or a recommendation indicating whether an applicant meets predefined criteria. In many cases, landlords receive only the summarized result rather than the underlying data.
Tenant screening algorithms are designed to estimate risk, often related to nonpayment of rent or lease violations. These systems apply weighted factors to different data points to generate a score or decision-support output.
Automated screening can streamline application review, but researchers and regulators have raised concerns about transparency, data accuracy, and whether certain data sources are appropriate predictors of rental outcomes.
These companies aggregate information from a range of public and private records to produce a consolidated profile used for screening decisions. Typical data inputs include:
Credit Reports and Credit Scores
Credit scores are widely used in financial assessments but were originally developed to predict loan repayment rather than rental payment behavior.Eviction Court Records
Screening reports may include eviction filings as well as case outcomes, depending on the data source.Criminal Background Data
Criminal history information is commonly included in screening reports.Income and Employment Verification
Verification is used to assess an applicant’s ability to meet rental obligations.Rental Payment History
While potentially informative, this data can be difficult to obtain and verify consistently.Public Records
This category can include bankruptcies, liens, judgments, and other legal or financial records.
The Mechanics of Algorithmic Decisions: How Risk Is Calculated
Tenant screening algorithms are designed to estimate the likelihood of outcomes such as missed rent payments or lease violations. They rely on statistical models that assign weights to various data points in order to generate a predictive profile.
While automation can improve efficiency, regulators and researchers have identified limitations related to accuracy, transparency, and the potential for disparate outcomes when models rely on incomplete or poorly contextualized data.
The SafeRent Lawsuit: Documented Algorithmic Limitations
The Louis v. SafeRent case provided insight into how automated screening systems may produce unintended outcomes.
Findings From the SafeRent Case
Court filings and settlement terms highlighted several documented issues:
Housing vouchers
The algorithm did not consistently account for the guaranteed income provided by housing vouchers when assessing ability to pay.Credit scores
Credit scores were heavily weighted despite mixed evidence regarding their ability to predict rental payment behavior.Eviction records
Eviction filings were included even when cases had been dismissed, resolved, or decided in favor of tenants, without consistent differentiation.
The Department of Justice and the Department of Housing and Urban Development filed a Statement of Interest emphasizing that the Fair Housing Act applies to automated systems when their outcomes result in disparate impact on protected classes.
Predictive Accuracy and Model Limitations
Tenant screening models are often based on the assumption that past behavior predicts future outcomes. However, several limitations are commonly discussed:
Credit scores and rental behavior
Credit scores were designed to predict loan repayment, not rent payment.Eviction filings and outcomes
Eviction filings may be treated similarly to completed evictions even when cases are dismissed, withdrawn, or resolved without a judgment.Context and individual circumstances
Automated models may not account for life events, temporary hardship, or long-term positive rental history.
Regulatory and Legal Context
Federal law does not prohibit the use of algorithms in housing decisions. However, existing statutes still apply, including:
- The Fair Housing Act, which prohibits practices that result in unjustified discriminatory effects
- The Fair Credit Reporting Act, which governs accuracy, disclosure, and dispute rights for consumer reports
In May 2023, HUD issued guidance encouraging housing providers to ensure that automated tools are used in a manner consistent with fair housing obligations, including appropriate human review.
Guidance documents are advisory and do not carry the force of law, but they reflect regulators’ interpretation of existing requirements.
Data Accuracy and Eviction Records
Eviction records play a significant role in tenant screening, but their reliability has been widely questioned.
Common issues include:
- Filings remaining on record after cases are dismissed or settled
- Records not updated after sealing or expungement
- Incorrect matches caused by similar names or incomplete identifiers
- Filings during COVID-era moratoriums that did not result in removals
In 2023, TransUnion agreed to pay 23 million dollars to resolve Consumer Financial Protection Bureau charges related to inaccuracies in tenant screening reports.
Criminal Background Data in Housing
Criminal background checks are frequently incorporated into tenant screening reports.
Concerns documented by housing researchers include:
- Inclusion of arrests that did not result in convictions
- Use of decades-old records without relevance to housing risk
- Limited consideration of rehabilitation or time elapsed
Landlord Considerations: Balancing Risk and Responsibility
Landlords face operational and financial risks, including unpaid rent, property damage, and legal costs.
Court cases and regulatory actions suggest that reliance on automated summaries without review of underlying data may introduce compliance and accuracy concerns.
Landlords using tenant screening services are generally encouraged to:
- Understand the screening criteria used by their chosen service
- Review underlying report data when possible
- Ensure compliance with Fair Credit Reporting Act and Fair Housing Act requirements
- Maintain a process for human review of borderline or unclear results
The Evolution of AI in Screening
As automated tools become more common, ongoing evaluation of data quality, transparency, and compliance remains important.
Transparency and Consumer Awareness
Applicants are often unaware of which company evaluated their rental application or what information was used.
Under the Fair Credit Reporting Act, applicants denied housing based on screening reports are entitled to receive an adverse action notice.
What Renters Can Do
Tenant screening can be difficult to navigate without visibility into the report used for a decision. Steps that may help include:
- Request tenant screening reports from the company used by the landlord.
- Review reports carefully for errors or outdated information.
- Dispute inaccuracies in writing.
- Request adverse action notices when denied.
- Prepare documentation explaining resolved records if applicable.
Why This Matters
Automated screening tools play an increasing role in housing decisions.
Understanding how these systems operate can help individuals better navigate the rental process.
Sources
- Louis v. SafeRent Solutions, LLC, Settlement Agreement (November 2024)
- DOJ and HUD Statement of Interest in Louis v. SafeRent (June 2022)
- HUD Guidance on Tenant Screening and Fair Housing (May 2023)
- CFPB v. TransUnion, Settlement (2023)