Source shares data on 1,803 US taxpayer-funded orgs using Clearview AI; many cops used its facial recognition without higher-ups knowing or after a local ban
A controversial facial recognition tool designed for policing has been quietly deployed across the country with little to no public oversight.
Context & Ripple Effects
BuzzFeed News' data release quantifies what early reporting on Clearview AI's investor and client 'testing' only hinted at: the tool spread through US government work far beyond any formal procurement. The 1,803-agency tally shows employees, not institutions, adopted it — which is why departments themselves often didn't know.
That bottom-up adoption pattern is what makes the follow-on coverage coherent: the GAO's audit missed five agencies' usage entirely, and by 2022 the FBI and ICE held formal contracts while Clearview pitched private companies too. The story arc is informal sprawl hardening into official use before oversight catches up.
First-order effects
- Individual officers at the named agencies ran ~340K searches without their departments' knowledge, and some continued after local bans — meaning existing municipal controls were already being bypassed at the user level.
- Agency leadership at the 1,803 organizations now faces an accountability gap: untracked tool usage that their own procurement and records processes never authorized.
Second-order effects
- Oversight bodies like the GAO are forced to rebuild their accounting from vendor-side data rather than agency self-reporting — the five overlooked agencies show official surveys undercount where adoption was informal.
- Vendors gain leverage from the sprawl: Clearview's later moves into formal government contracts and private-company sales show a bottom-up user base being converted into an institutional customer base.
Third-order effects
- If informal employee adoption keeps preceding formal review, facial recognition governance shifts from deciding whether to adopt to retroactively regulating tools already embedded in police work — pushing toward procurement observability requirements and vendor-side audit trails.
- The ban-versus-usage gap points toward regulation aimed at data sourcing and vendor conduct rather than local adoption rules, since municipal prohibitions demonstrably failed to stop usage.
The trend: Government AI tools are spreading through individual employee adoption faster than institutional procurement and oversight can track, making vendor-side records the de facto audit layer.