Investigation: some US public housing agencies use surveillance cameras with facial recognition and other AI to evict residents, sometimes for minor violations
Surveillance cameras purchased with federal crime-fighting grants are being used to punish and evict public housing residents …
Context & Ripple Effects
The oversight vacuum here was documented years ago: when Sen. Wyden asked in 2020, HUD said it neither monitors nor tracks facial recognition use in public housing, leaving agencies free to deploy grant-funded cameras without federal accounting of what the systems do. This investigation supplies the missing answer — the cameras bought for crime-fighting are also running facial recognition and other AI against residents themselves.
The story extends an arc the coverage has tracked since city-wide AI video networks began funneling private feeds to police HQs: surveillance purchased as a policing tool quietly becoming an administrative one. It converges with the parallel finding that landlords are adopting opaque AI tenant-screening tools despite warnings about errors and discrimination — housing is emerging as the sector where algorithmic judgment decides who keeps a roof.
First-order effects
- Residents of the affected housing agencies face eviction triggered by camera-flagged minor violations, turning federal anti-crime equipment into a lease-enforcement instrument aimed at the tenants who live under it.
- HUD's 2020 stance that it does not track facial recognition in public housing is now directly contradicted by documented agency deployments, putting the department's hands-off posture on record as a governance failure.
Second-order effects
- Federal crime-fighting grant programs face pressure to attach usage conditions and reporting requirements, since the same dollars funding police cameras are demonstrably financing tenant surveillance.
- Local ordinances restricting government facial recognition — the kind New Orleans violated while secretly running 200+ street scanners per the related investigation — become the model privacy advocates will push housing agencies and city councils to adopt.
Third-order effects
- If the pattern holds, surveillance becomes a structural condition of subsidized housing itself: infrastructure justified as crime prevention normalizes into routine eligibility and compliance enforcement, widening the gap between what agencies buy and what any regulator tracks.
- Housing joins policing as a domain where identification systems act on people with no corroborating human review — echoing cases of suspects arrested on facial recognition matches alone — and forcing the question of whether civil-housing law, not just criminal-law debates, needs its own rules for automated identification.
The trend: Surveillance infrastructure bought with public-safety money is migrating from policing into housing administration faster than any federal or municipal regulator tracks its use.