Los Angeles scoring system ranks Black and Latino unhoused people lower for subsidized housing

Investigations by The Markup and the Los Angeles Times reported that a scoring system used to prioritize unhoused people for subsidized permanent housing in Los Angeles produced consistently lower priority scores for Black and Latino people. The reporting analysed intake assessment records and found these disparities persisted year after year, making Black and Latino people less likely to receive permanent housing. Subsequent reporting says the city and local agencies moved to change how vulnerability is scored.

City of Los Angeles; Los Angeles Homeless Services Authority (LAHSA) · Incident Feb 28, 2023 · Indexed Jun 10, 2026 · 3 sources

The vulnerability scoring algorithm assigned systematically lower priority scores to Black and Latino people, reducing their chances for housing.
What
Investigations by The Markup and the Los Angeles Times reported that a scoring system used to prioritize unhoused people for subsidized permanent housing in Los Angeles produced consistently lower priority scores for Black and Latino people.
Incident date
Feb 28, 2023
Who
City of Los Angeles; Los Angeles Homeless Services Authority (LAHSA)
Failure mode
Policy Violation
AI surface
Algorithmic Decision
Severity
High

What happened

In February 2023 The Markup and the Los Angeles Times published an investigation that analysed intake assessment records and found the scoring system used to prioritise unhoused people for subsidized permanent housing produced lower priority scores for Black and Latino people. The system has been used to decide who receives limited permanent housing resources, and the reported disparities meant Black and Latino individuals were less likely to be placed into housing. Subsequent reporting documented that Los Angeles and its homelessness agencies said they would change how they score vulnerability.

What broke inside the model

Failure path · mode profile · Policy Violation
  1. 01 · TriggerA prompt pushes against a deployment boundary.
  2. 02 · Model stepThe model produces the disallowed output.
  3. 03 · Control gapNo enforcement blocks it at generation time.
  4. 04 · FailureThe output crosses the policy line.
  5. 05 · ConsequenceA limit the business set is breached in public.

The output crosses a policy boundary the deployment had defined.

The failure was an algorithmic fairness problem: the scoring algorithm's design, including the assessment questions and weightings, produced outputs that correlated with race and produced a disparate impact. Prior research and reporting have flagged the commonly used vulnerability assessment (VI-SPDAT and related implementations) for similar biases, and L.A.'s implementation and weighting choices amplified those disparities. This is a systemic bias in the scoring workflow that led to inequitable prioritization rather than a technical outage.

Public visibilityHigh
Regulatory exposurePossible
Customer impactMany customers
Financial impactUnknown
Time to disclosureMonths
  1. PressL.A.'s Scoring System for Subsidized Housing Gives Black and Latino People Experiencing Homelessness Lower Priority Scoresthemarkup.org
  2. PressBlack and Latino homeless people ranked lower on priority listlatimes.com
  3. PressL.A. Is Changing How It Scores “Vulnerability” of Unhoused Peoplethemarkup.org
Permalinkhttps://failureindex.ai/failures/los-angeles-scoring-ranks-black-latino
CitationAI Failure Index. "Los Angeles scoring system ranks Black and Latino unhoused people lower for subsidized housing" (FI-0399). Realm Labs. https://failureindex.ai/failures/los-angeles-scoring-ranks-black-latino (indexed Jun 10, 2026).
Share cardA branded image of this record for posts and slides.

Data fields CC-BY 4.0, prose citation permitted. Incident ID FI-0399. Full dataset at /data.

Note from Realm Labs, the Index steward

How Realm fits

Controls for this failure mode
  • Prism
  • OmniGuard

This entry sits in the index's predictive wing: a system that scores, ranks, perceives, or steers rather than generates. Realm's runtime layer is built for the generative and agentic systems now moving into these same decision seats, where it watches a model's internal state and holds an unsupported claim or an unchecked action before it commits. The control gap on this record, an automated decision that reached people with no runtime check in front of it, is the same gap. The index keeps predictive failures on the record because the pattern carries straight into the systems shipping today.