Robert Dillon’s Lawsuit Turns a Florida Arrest Into a Test of Police Facial Recognition
Robert Dillon’s lawsuit says Florida police treated a 93% facial-recognition hit like a case file, not a lead. The filing is now a sharp test of how courts may judge police use of AI when basic checking should have come first.
Robert Dillon’s federal lawsuit does not read like a warning about some distant AI future. It reads like a warning about ordinary police work that stopped being ordinary the moment software was allowed to do the heavy lifting. In a complaint filed on June 10, 2026, Dillon says officers in northeast Florida treated a facial-recognition result as the spine of a child-luring case even though he lived more than 300 miles away in Fort Myers, said he had never been to Jacksonville Beach, and could have been ruled out by routine checks.
First Coast News — Fort Myers man sues Jacksonville Beach police, alleges false arrest tied to facial recognition tech
First Coast News' local report tracks the June 12 lawsuit update; if the player does not load, use the direct YouTube link in the article body.
The filing, backed by the ACLU and ACLU of Florida, matters because it shifts the argument away from whether facial recognition is impressive technology and back to a simpler question: what are officers supposed to do after an algorithm points at a face? Dillon’s answer is that they are supposed to investigate. His lawsuit says they did the opposite.
Watch: First Coast News’ report on Dillon’s lawsuit and the arrest record on YouTube.
The claim is not just that the software was wrong
According to the complaint, Jacksonville Beach police sought a warrant on July 31, 2024, and a judge signed it on August 1, 2024, after officers relied on a Jacksonville Sheriff’s Office facial-recognition search that described Dillon as a 93% match. The complaint argues that this number looked precise while hiding the real weakness in the process. A similarity score, it says, is not the same thing as a probability that two images show the same person.
That distinction is central to the suit. The filing says the probe image came from poor-quality surveillance material and that the officers still moved forward while leaving out exculpatory details, including license-plate reader data that allegedly showed Dillon’s vehicles were not in Duval County around the incident, plus the obvious geographic problem that a Fort Myers resident was being treated as a supposed regular at a Jacksonville Beach McDonald’s.
“If you came to me with a facial recognition hit and that was your probable cause, I would probably kick you out of my office because that’s not how it works.”
Jacksonville Sheriff T.K. Waters, as quoted in Dillon’s complaint
That quote is useful because it captures the case’s political danger for law enforcement. Even if police departments want facial recognition to remain an investigative tool, the lawsuit tries to show how quickly a tool becomes a shortcut once a number arrives wrapped in technical authority.
What the lawsuit says police had, and what it says they ignored
| What investigators had | Why Dillon says it was not enough |
|---|---|
| A facial-recognition result listing Dillon as a 93% match | The complaint says the score measured image similarity, not the likelihood that Dillon was the suspect. |
| A witness photo-array identification | The lawsuit says the lineup was already tainted because Dillon entered it only after the algorithm flagged him. |
| An arrest-warrant narrative built around the match | The complaint says officers omitted exculpatory facts, including Dillon’s distance from Jacksonville Beach and his denial. |
| Months between the incident and the warrant request | Dillon argues that time should have allowed for basic corroboration rather than a rush to confirm the software. |
The broader point is hard to miss: once Dillon’s face was pulled into the system, the complaint says every later step looked less like independent verification and more like reinforcement of the same original mistake.
Why this case reaches beyond one Florida town
The complaint describes FACES, the Pinellas County system used in the search, as one of the oldest and widest-reaching law-enforcement facial-recognition databases in the country. It says that as of 2022 the system held more than 38.5 million images, was available to at least 196 agencies, and processed more than 8,000 searches a month. That scale is exactly why Dillon’s case matters outside Jacksonville Beach. A bad arrest tied to a small local department can still expose a statewide habit.
The filing also places Dillon inside a larger national pattern. It says he is one of at least 15 publicly known people in the United States whose arrests were tied to faulty police reliance on facial recognition. That does not prove every system works the same way, but it does make one defense much harder to sustain: that these cases are too isolated to require new rules.
A short timeline of the case
- January 2024: The underlying child-luring investigation begins in Jacksonville Beach, according to the complaint.
- July 31, 2024: Officer Scott O’Connell submits the warrant request, the filing says.
- August 1, 2024: A judge signs the arrest warrant, according to the complaint.
- August 2024: Dillon is arrested at his home by a Lee County deputy, held overnight, and posts bond by borrowing money and pledging his truck title, the suit says.
- October 7, 2024: Dillon pleads not guilty.
- October 24, 2024: Prosecutors drop the charges, 69 days after the wrongful arrest, according to the complaint.
- June 10, 2026: Dillon files his federal civil-rights lawsuit.
That sequence is what makes the story bigger than a headline about AI gone wrong. The alleged damage was not confined to one arrest. The complaint says Dillon spent months living under the stigma of a child-luring charge, saw his mugshot remain online, and now avoids normal interactions around children because of what the accusation did to his reputation.
What to watch next
The legal fight will now turn on a mix of old constitutional questions and newer technical ones: whether officers omitted key facts from the warrant application, whether the lineup can be treated as independent corroboration, and whether agencies using facial-recognition systems can keep calling them mere leads when the paperwork shows the match driving the case.
For readers outside Florida, the useful takeaway is less dramatic and more practical. The next policy battle over police AI will probably not be about whether a department can buy a system. It will be about whether courts, lawmakers and police supervisors force rules on what must happen before an algorithmic match can ever get near handcuffs.
Primary documents and reporting used for this article include the federal complaint in Dillon v. City of Jacksonville Beach, the ACLU of Florida’s case release at ACLU.org, and local reporting from First Coast News.
Read Next
Related Stories
Anthropic’s Fable Shutdown Turns AI Export Control Into a Product-Trust Test
Anthropic says a U.S. export-control directive forced it to suspend Fable 5 and Mythos 5 just days after launch. The bigger technology story is not only what the government feared, but how fragile frontier-model access now looks to customers, researchers and rivals.
Japan's H3 Rocket Returned to Flight. The Real Test Was Whether Low Cost Had Survived Failure.
JAXA's June 12, 2026 H3 launch mattered less as a routine mission than as proof that Japan's cheaper rocket configuration still had a commercial future after the program's December 2025 failure.
NOAA Says El Niño Has Arrived. The Bigger Signal Is How Fast Uncertainty Is Narrowing.
NOAA's June 11, 2026 El Niño advisory matters less because the label became official and more because the forecast window got sharper: forecasters now see a 63% chance of a very strong event by winter, which changes how governments, utilities and food markets should read the rest of the year.