A Texas jury priced an AI dragnet at $76 — the Fourth Amendment won, four years late and for pocket change

A federal jury on Monday found Bexar County, Texas, liable for systematically violating thousands of drivers' Fourth Amendment rights through its criminal interdiction unit, in a case brought by the Institute for Justice, a nonprofit law firm. The chain that produced those violations started with software: a nationwide network of AI-powered license plate readers that, according to the lawsuit, flagged travel patterns deemed "abnormal" and notified local law enforcement.

The facts, as laid out in the complaint: on March 16, 2022, Alek Schott was driving home from a work trip near San Antonio when Bexar County Sheriff's Office Deputy Joel Babb stopped him for allegedly drifting lanes. Babb asked him to sit in the cruiser's passenger seat, then questioned him for roughly ten minutes about his job, his travel and his home while pulling his criminal history. The suit says Babb conceded the stop was never really about a traffic violation — he didn't write tickets, he looked for "human smuggling, drug smuggling, and all those things." Schott refused consent to a search. A K-9 unit was called; body camera footage shows the handler gesturing upward, the dog jumping onto the truck, and a positive alert being recorded. Officers searched for 40 minutes and found nothing. Schott's own dashcam later showed he never drifted lanes, the suit says. When he complained, internal affairs told him that if he didn't like how they did business, he should file a lawsuit.

The Institute for Justice describes a five-step pipeline: plate readers flag anomalous travel, deputies make pretextual stops, drivers are interrogated, and if consent is refused, a manipulated dog alert supplies the probable cause. Senior attorney Josh Windham put the logic plainly: "It all rests on the idea that if you watch enough innocent people closely enough, criminals will fall out." Lead counsel Christie Hebert said officers have to actually believe a crime is being committed, and "these officers didn't have that." The jury awarded Schott $76 — one dollar for every minute he was unconstitutionally detained.

This is where the technology question gets sharper than "AI made a mistake." The readers didn't arrest anyone. What they did was make suspicion nearly free. Surveillance used to be rationed by cost: following a car took an officer and a shift. A camera network removes that budget constraint, and any classifier aimed at millions of lawful drivers will generate overwhelmingly innocent hits — not because it's badly built, but because crime is rare and "abnormal" is not a crime. The scandal described in this suit is what happened next: a weak statistical signal was laundered into legal probable cause through a dog that, per the footage cited in the complaint, was cued.

Our read: the constitutional system worked — and the terms of that victory should worry everyone. Four and a half years, a specialist nonprofit litigating full-time, and a verdict that prices thousands of invasive stops at $76 for the man who fought back. Accountability set far below the cost of the behavior it's meant to deter is not deterrence; it's a receipt. The practical fixes are unglamorous and available today: mandatory logging of which machine alert triggered which stop, independent audits of hit-to-conviction rates, and a bright-line rule that an algorithmic flag can never be converted into probable cause by a handler's body language.

And this matters beyond policing. We hold a long-term view that AI's payoff is medical — diseases erased, healthspan extended, abundance that lets people work on what they love. That future runs on public consent, and consent is spent every time AI's most visible civic appearance is a dragnet that treats ordinary travel as evidence. Bexar County didn't just lose a lawsuit. It made the case, one more time, that the hardest problem in deploying AI isn't the model — it's the institution holding it.