Sources: Palantir has been secretly testing predictive policing technology in New Orleans since 2012 that identifies individuals at risk of committing crime
Palantir deployed a predictive policing system in New Orleans that even city council members don't know about — Illustrations by Garret Beard and Alex Castro
Context & Ripple Effects
This report lands on top of Wired's 2017 investigation into how Palantir built an opaque surveillance network by pooling data from police departments and government agencies while locking in clients despite complaints about low transparency. The New Orleans deployment is that playbook in miniature: a person-level risk-scoring system running inside the police department since 2012, hidden even from city council members who nominally oversee it.
It also sits within a broader predictive-policing wave — PredPol, which grew out of an LAPD/UCLA research project, had already spread to more than 60 jurisdictions by 2016 — but Palantir's version is distinguished by its secrecy and its focus on flagging individuals rather than places.
First-order effects
- New Orleans city council members are confronted with a multi-year police technology program they were never told about, forcing an immediate accountability question for the department and its vendor.
- Palantir faces its first public test of whether its Gotham-style law-enforcement business can survive disclosure — the same software whose FOIA-released user manual later showed how much information officers can pull up on any individual.
Second-order effects
- Other cities running Palantir systems come under pressure to disclose their own contracts, threatening the quiet renewal model that made New Orleans possible.
- Transparency complaints give city officials grounds to walk away — pressure that culminates when New Orleans announces it will not renew the Palantir contract.
Third-order effects
- If the pattern holds, secret deployments recur after each reckoning: years later, an investigation finds New Orleans police ran 200+ facial recognition cameras for two years, apparently violating a 2022 ordinance — suggesting procurement secrecy outlasts any single contract fight.
- Bias findings against place-based tools like PredPol, which investigations found steered police toward poor, Black, and Latino neighborhoods, push the structural debate toward whether person-level prediction should be governed at all.
The trend: Predictive policing is migrating from visible pilot programs to covert, individual-level surveillance systems, with municipal oversight repeatedly arriving only after journalists force disclosure.