Florida man sues police over wrongful arrest based on faulty facial recognition match

At a glance:

• Robert Dillon was arrested in August 2024 after a facial recognition system incorrectly matched him as a suspect in a child-luring case.
• The lawsuit claims police used a 93% confidence score from FACES without proper verification and concealed exculpatory evidence.
• Dillon, 52, from Fort Myers, was held for over two months before charges were dropped, causing significant personal and professional harm.

What happened

Robert Dillon, a 52-year-old Fort Myers resident, filed a lawsuit against Florida law enforcement officials in the US District Court for the Middle District of Florida. The case stems from his arrest in August 2024 on charges of attempting to lure a child under twelve at a Jacksonville Beach McDonald’s. According to the complaint, a facial recognition system called the Face Analysis Comparison and Examination System (FACES) flagged Dillon as a 93% match to a suspect captured in low-quality surveillance footage. However, Dillon had never been to Jacksonville Beach, over 300 miles from his home, and license plate reader data confirmed his vehicles were not in the area during the alleged incident. Despite this, police proceeded with the arrest, relying solely on the algorithmic match without further investigation.

The arrest occurred at Dillon’s home in front of his wife, who immediately informed officers he had never visited Jacksonville Beach. Dillon himself told deputies he hadn’t left Fort Myers in two years. He was held overnight in jail, forced to pledge his truck as collateral for bond, and faced public stigma after his mugshot was published online. Charges were dropped after over two months, but the lawsuit alleges that officers deliberately ignored exculpatory evidence, including Dillon’s alibi and physical characteristics that didn’t match the suspect.

How facial recognition failed

The lawsuit highlights critical flaws in the FACES system, a centralized facial recognition database maintained by the Pinellas County Sheriff’s Office. As of 2022, FACES contained over 38.5 million images and was accessible to at least 196 law enforcement agencies. The 93% confidence score cited in the case was derived from a photo of a McDonald’s computer screen displaying surveillance footage—a method that introduced additional degradation due to screen glare, reduced resolution, and color distortion. The plaintiffs argue that such scores are not probabilistic measures but rather indicators of digital proximity between facial templates, making them unreliable for identifying individuals.

The system’s limitations were compounded by procedural errors. Officer Scott O’Connell, who conducted the FACES search, presented a photo array to a McDonald’s manager that included Dillon and five fillers designed to resemble him, not the actual suspect. This setup increased the likelihood of a false positive. Furthermore, the affidavit for the arrest warrant omitted key information, such as the unreliability of facial recognition results under Jacksonville Beach PD’s own policy and the lack of corroborating evidence. The lawsuit emphasizes that officers had no framework to evaluate the validity of the 93% score, leaving them dependent on a flawed technology.

Investigation shortcomings

The complaint alleges that O’Connell and other officers actively avoided verifying the facial recognition match. Despite having access to mobile ordering records, payment data, McDonald’s app accounts, and cell phone location data, they failed to request these records. A McDonald’s manager’s description of the suspect as a “regular customer” was misrepresented in the affidavit, implying she witnessed the incident when she was actually occupied with work duties. O’Connell also omitted a prior phone call with Dillon, during which he denied involvement and described a distinctive scar that did not appear in the surveillance footage.

The lawsuit further criticizes O’Connell’s suitability for the case, citing his documented history of volatility and poor judgment. He had previously been terminated from the St. Johns County Sheriff’s Office for threatening to “blow up” the agency, later reinstated, and then arrested for domestic battery before resigning. Despite this record, Jacksonville Beach PD hired him and later promoted him to corporal after the wrongful arrest. The complaint argues that these decisions reflect systemic negligence in deploying officers with questionable judgment on sensitive cases involving children.

Impact on the plaintiff

Dillon’s arrest had severe personal and professional repercussions. As a self-employed commercial crabber, he missed a critical month of work during a lucrative season, falling behind on rent and fearing eviction. The public availability of his mugshot led to ongoing harassment, with community members approaching him to inquire about the case. He reported feeling unable to interact comfortably with children and expressed lasting trauma, stating he “will never get over how terrified and worried I was, wondering if I’d ever go home to my wife and daughter again.”

The lawsuit seeks financial damages and policy reforms to prevent future misuse of facial recognition technology. It underscores the broader risks of relying on AI systems without proper oversight, particularly in cases involving serious criminal charges. The ACLU, representing Dillon, noted he is one of 15 known individuals in the US to have been wrongly arrested due to facial recognition errors. The case raises urgent questions about transparency, accountability, and the ethical use of biometric data in law enforcement.

Broader implications

The lawsuit reflects growing concerns about facial recognition’s reliability and its potential for abuse in policing. While the technology is marketed as a tool for solving crimes, the case illustrates how it can lead to wrongful arrests when not paired with rigorous human verification. The FACES system’s widespread use across 196 agencies highlights the need for standardized protocols to assess algorithmic accuracy and prevent overreliance on automated results.

Experts have long warned that facial recognition systems perform inconsistently across demographics and are prone to errors with low-quality images. The lawsuit’s focus on the 93% score’s misinterpretation adds to calls for clearer guidelines on how law enforcement interprets such metrics. As legislators grapple with regulating biometric surveillance, cases like Dillon’s may influence future policies on transparency, training, and accountability in AI-driven investigations.

Legal and technical context

The complaint emphasizes that facial recognition results are inherently unreliable and cannot constitute probable cause under Jacksonville Beach PD’s own policy. However, the affidavit for Dillon’s arrest warrant failed to disclose this limitation, misleading the magistrate who approved the arrest. The lawsuit argues that officers have a duty to independently verify algorithmic matches, especially in cases involving severe charges like child endangerment.

The technical challenges of facial recognition are well-documented. Confidence scores like the 93% cited here are not standardized across systems, making them difficult to interpret. The lawsuit’s critique of FACES aligns with studies showing that even high scores can result in false positives when images are of poor quality or when algorithms are trained on biased datasets. These issues underscore the need for clearer legal frameworks governing the use of biometric data in criminal investigations.

What to watch next

The lawsuit’s outcome could set a precedent for how facial recognition evidence is handled in court. Legal experts will likely scrutinize whether the plaintiffs can prove malicious prosecution and whether the defendants’ actions violated constitutional protections against unreasonable searches and seizures. The case may also prompt legislative action, as Florida and other states consider stricter regulations on law enforcement’s use of biometric technologies.

For now, the Jacksonville Sheriff’s Office and Jacksonville Beach Police have not responded to requests for comment. The ACLU and Dillon’s legal team are seeking systemic changes to prevent similar incidents, including mandatory training on facial recognition limitations and requirements for independent verification before arrests. As public scrutiny of AI in policing intensifies, this case highlights the urgent need for transparency and accountability in automated systems.