The NYPD has developed Patternizr, an AI-based system to track crimes and spot patterns using police data, and has been using it since late 2016
J. Brian Charles / Governing :
Context & Ripple Effects
Patternizr did not arrive with an announcement: by the time Governing reported on it, the NYPD had been running the system internally since late 2016, mining its own case data to surface crime patterns. That quiet deployment fits a documented lineage — [[a:867205|PredPol grew out of an LAPD/UCLA research project into a tool used across dozens of jurisdictions]], and Palantir spent years secretly testing predictive policing in New Orleans before that too surfaced through reporting.
What distinguishes the NYPD case is scale and integration: the department pairs Patternizr with a counterterrorism apparatus whose budget has more than quadrupled since 2006, and reporting shows it repurposing post-9/11 surveillance tools for everyday street crime — much of it collectible without a warrant. The NYPD's own data has also fed commercial systems before, as when IBM used NYPD surveillance footage in 2012 to develop object identification tech.
First-order effects
- NYPD detectives get machine-generated pattern matches across precincts, shifting how cases are linked and which suspects draw coordinated attention — with the department now answering for more than two years of undisclosed use.
- New Yorkers flagged by Patternizr-linked investigations face decisions made partly by a system they were never told existed, compounding the warrantless-collection practices already reported around the department's surveillance toolkit.
Second-order effects
- Vendors in the predictive-policing market — PredPol, Palantir, and the license-plate-recognition startups like Rekor and Flock selling 'suspicious movement' detection to local governments — gain proof that a flagship department will adopt and sustain these systems even under eventual scrutiny.
- Other departments can follow the NYPD's deploy-first-disclose-later template, lowering the political cost of adoption and accelerating procurement of pattern-analysis tools nationwide.
Third-order effects
- If the pattern holds, pattern-detection AI becomes default infrastructure in American policing rather than an experiment — pushing the accountability fight from whether departments adopt such systems toward auditing ones already embedded, as later coverage of body-cam analysis software like Truleo suggests is underway.
- Sustained backlash episodes like the Spot robot-dog controversy point toward formal governance regimes — disclosure requirements, audits, possibly regulation — becoming the mechanism that determines which surveillance tools survive public deployment.
The trend: Predictive policing is moving from research pilots and secret trials into quietly embedded, always-on infrastructure inside major police departments.