Robert Dillon, a resident of Fort Myers, is suing several Florida law enforcement agencies after being wrongfully arrested for a child-luring incident. The arrest occurred after a facial recognition algorithm incorrectly identified the 52-year-old man as a suspect in a case located in Jacksonville Beach.

How the 2001 FACES algorithm flagged a man 300 miles away

The wrongful arrest of Robert Dillon began with the Face Analysis Comparison and Examination System, known as FACES. According to the report, the Pinellas County Sheriff's Office has operated FACES since 2001, making it one of the oldest facial recognition tools used by police in the United States. At its peak in 2021, the system provided law enforcement across Florida with access to tens of millions of driver's license photos and mugshots.

In this instance, the FACES algorithm flagged Robert Dillon as a potential match for a suspect who allegedly approached a girl at a McDonald's in Jacksonville Beach shortly before midnight on November 2, 2023. However, Robert Dillon lived more than 300 miles away from the scene of the crime and had never visited the Jacksonville Beach area. Despite denying any involvement in December, Mr. Dillon was arrested at his home the following August.

Corporal Scott O'Connell's history of volatility and the arrest warrant

The legal action targets not only the technology but the human failures of the lead investigator, Scott O'Connell. As the report states, Officer O'Connell treated the output of the facial recognition algorithm as a near-certain identification rather than a lead. he is accused of omitting critical exculpatory evidence from the arrest warrant application and failing to perform basic investigative steps that would have cleared Robert Dillon immediately.

The lawsuit further highlights a troubling employment history for Scott O'Connell. He was previously terminated from the St. Johns County Sheriff's Office after threatening to "blow up" the agency. Although he was later reinstated, he eventually resigned following an arrest for domestic battery. Despite this record, the Jacksonville Beach Police Department hired him and later pormoted him to the rank of corporal after his investigation led to the wrongful prosecution of Mr. Dillon.

The danger of treating algorithmic matches as absolute proof in Florida

This case reflects a broader, systemic trend where law enforcement agencies may succumb to automation bias, treating a machine's "possible match" as an empirical fact. When officers build a case to confirm an algorithm's answer rather than testing that answer against evidence , the technology ceases to be a tool and becomes a catalyst for civil rights violations. The Technology Project has argued that Florida police must implement strict safeguards to prevent such failures from recurring.

For Robert Dillon, the consequences extended beyond a single night in jail. He was forced to pledge the title of his truck to post bond and suffered the lasting stigma of a public mugshot that remains online. The psychological toll has been significant; Mr. Dillon stated that the terror of wondering if he would ever return to his wife and daughter still haunts him, and he no longer feels comfortable interacting with strangers in public.

The missing identity of the November 2 suspect

While the lawsuit addresses the damage done to Robert Dillon, several critical questions remain unanswered. Most notably, the identity of the actual suspect who approached the child at the Jacksonville Beach McDonald's on November 2 remains unknown . The source does not indicate if the Jacksonville Beach Police Department has since identified the real perpetrator or if the case has gone cold due to the initial investigative failure.

Additionally,the report leaves open the question of why the Jacksonville Beach Police Department's hiring process failed to flag the severity of Scott O'Connell's prior misconduct. There is no information provided regarding whether other arrests were made based on O'Connell's reliance on the FACES system,suggesting a potential pattern of errors that has yet to be fully uncovered .