SafeRent Algorithmic Housing Discrimination Settlement (2024)

Incident Summary

SafeRent Solutions, a widely used tenant screening company, agreed to a $2.275 million class action settlement after its algorithmic screening system was alleged to disproportionately reject Black and Hispanic rental applicants who used housing vouchers.^[1]

The lawsuit alleged that the algorithm failed to account for the subsidy component of housing vouchers when evaluating applicants’ ability to pay rent. By over-weighting credit scores and not adjusting for guaranteed government subsidies, the system allegedly produced scores that disadvantaged applicants from racial and ethnic groups that disproportionately rely on housing vouchers.

The case resolved via class action settlement, without a court determination on liability. The settlement required algorithmic modifications and represents one of the first class actions challenging algorithmic discrimination in housing.

Key Facts

• Settlement amount: $2.275 million
• Defendant: SafeRent Solutions (tenant screening company)
• Alleged mechanism: Algorithm over-weighted credit scores and failed to account for housing voucher subsidies
• Affected populations: Black and Hispanic rental applicants using housing vouchers
• Outcome: Class action settlement with required algorithmic modifications; no admission of liability

Threat Patterns Involved

Primary: Allocational Harm — The lawsuit alleged that the tenant screening algorithm denied housing opportunities to qualified applicants from protected groups by failing to account for voucher subsidies in its scoring model.

Significance

This incident is significant for several reasons:

1. Housing as high-stakes allocation — Unlike many algorithmic bias cases in hiring or lending, this case concerns access to housing, a fundamental need where algorithmic denial can have immediate and severe consequences.
2. Structural design flaw — The alleged discrimination did not arise from biased training data alone but from a fundamental design failure: the algorithm allegedly did not account for a well-known government subsidy program.
3. Precedent for tenant screening — As AI-powered tenant screening becomes standard practice, this settlement signals that vendors may face liability for discriminatory outcomes in housing decisions.
4. Proxy discrimination — The case illustrates how facially neutral variables like credit scores can serve as proxies for race when correlated with socioeconomic factors that themselves reflect historical discrimination.