A 52-year-old Florida man says an algorithm turned his life upside down—before any real investigation even began.
A “93% facial recognition match” was enough to trigger an arrest in a child-luring case he insists he had nothing to do with.
Now he’s fighting back in federal court, and the case is raising a hard question: how much power should AI have over police decisions?
Table of Contents
ToggleWhat Happened
In August 2024, Robert Dillon of Fort Myers was arrested over an alleged attempt to lure a child at a McDonald’s in Jacksonville Beach.
The key trigger wasn’t a traditional lineup or direct eyewitness ID—it was a facial recognition hit from a system linked to the Pinellas County Sheriff’s Office database called the Face Analysis Comparison and Examination System (FACES).
That system reportedly flagged Dillon as a 93% match to a suspect captured on low-quality surveillance footage—actually a photo of a computer screen showing the video, not the original file.
Police then moved forward with the case.
But Dillon’s lawsuit, supported by the American Civil Liberties Union, claims something far more troubling happened next:
- License plate data showed no sign his vehicles were in the area
- He lived over 300 miles away
- He had never been to Jacksonville Beach
- Even a police phone call with him revealed denial and a visible facial scar not matching the suspect
Still, the arrest went ahead.
Why It Matters
At the center of the case is a single number: 93%.
To most people, that sounds almost certain. But the lawsuit argues it’s misleading.
The system behind it—FACES, used by at least 196 law enforcement agencies and containing 38.5 million images—does not produce certainty. It produces a mathematical similarity score.
And that distinction is now at the heart of the legal battle.
Key concern raised in the lawsuit:
Officers treated a confidence score like a positive identification.
Dillon was arrested, jailed overnight, and charged with a deeply stigmatizing offense—before the case was eventually dropped about two months later.
But by then, the damage was already done: mugshots online, public suspicion, and months of legal stress.
Must Read: 5 Scientists Ejected in 2026 ADA Scandal Sparks Shocked Apology
Hidden Problem Inside the Investigation
The lawsuit doesn’t just blame AI. It also accuses investigators of ignoring clear evidence.
According to the filing:
- A license plate reader search showed no movement into Duval County
- McDonald’s ordering and payment records were never requested
- Phone, travel, and location data were not pursued
- A manager’s statement was misrepresented as eyewitness identification
Even more sharply, the complaint says the investigator:
“Did not merely fail to investigate… but chose not to pursue readily available evidence.”
The surveillance image itself was also heavily degraded—taken from a screen display of a video, adding glare, distortion, and loss of detail before it even entered the AI system.
The Human Fallout
Beyond the legal arguments, Dillon’s life changed immediately.
- Held overnight in jail
- Forced to borrow money and pledge his truck for bond
- Lost about a month of work as a commercial crabber
- Faced ongoing public stigma after charges were dropped
He now says strangers still question him in public—and that he avoids normal interactions out of fear of being misjudged.
One quote from Dillon stands out:
“I will never get over how terrified and worried I was, wondering if I’d ever go home to my wife and daughter again.”
Contrarian View: Was AI Really the Problem?
Not everyone agrees the technology itself is the issue.
Supporters of facial recognition argue:
- AI is not supposed to replace investigation—only guide it
- Human officers made the final decisions
- The real failure may lie in how the system was used, not the system itself
In that framing, the 93% match wasn’t the problem—it was the decision to treat it as conclusive evidence without stronger verification.
This raises a difficult question:
If a tool is known to be imperfect, who is responsible when it is still used as certainty?
What Happens Next
The lawsuit targets multiple agencies and officials, including:
- Jacksonville Beach Police leadership
- Jacksonville Sheriff’s Office personnel
- Supervisors tied to FACES database usage
Dillon is seeking damages and changes to how facial recognition evidence is used in investigations.
The case also highlights a broader national issue: many agencies still lack clear legal boundaries on how AI-generated matches can be used in warrants or arrests.
The Bigger Question Ahead
This case may ultimately hinge on something bigger than one arrest: whether courts decide that a facial recognition score can ever be treated as reliable enough to trigger criminal charges.
Because if a “93% match” can lead to an arrest today, the question becomes what happens when similar systems become faster, cheaper—and even more widely used.
And that leaves one unresolved tension hanging over the entire case:
When AI points to a suspect, how much doubt is enough to stop the handcuffs from coming out?
Disclaimer: This article is based on publicly available reporting and court filings. No facts, timelines, or outcomes have been invented. Interpretations reflect analysis and may evolve as new information emerges.