A false arrest lawsuit filed this week in Florida alleges that police relied on a flawed facial recognition match and then concealed evidence that would have cleared the suspect. The plaintiff, Robert Dillon, was arrested in August 2024 after a facial recognition system flagged him as a 93 percent match to a person seen on a McDonald's surveillance camera in Jacksonville Beach. Dillon lives more than 300 miles away in Fort Myers and has never visited that area, according to the complaint.
The lawsuit, filed against the Jacksonville Sheriff's Office and several officers, claims investigators built a case to confirm the algorithm's output rather than test it against available evidence. The arrest was based on a low-quality image taken from a photo of a computer screen displaying video footage. A search of license plate reader databases found no record of Dillon being near Jacksonville Beach at the time of the alleged incident.
The Case Against the Algorithm
Facial recognition technology has become a common tool for law enforcement across the United States. Supporters argue it speeds up investigations and helps identify suspects from security footage. But critics say the systems are prone to error, especially with poor-quality images or when scanning faces of people of color. Studies have shown that many commercial facial recognition algorithms produce false positives at significantly higher rates for Black and Hispanic individuals compared to white individuals.
In Dillon's case, the image used for comparison was a screen capture of a surveillance monitor, introducing grain and distortion that likely degraded the match. The 93 percent confidence figure is not a guarantee of accuracy. Many systems operate with thresholds that can generate false matches even at high confidence levels, particularly when the input image is compromised.
Alleged Evidence Suppression
The lawsuit goes beyond the initial algorithmic error. It alleges that officers deliberately ignored or concealed exculpatory evidence after the arrest. Dillon was charged with attempting to lure or entice a child under 12 years old. Prosecutors later dropped the charges, but Dillon spent time in jail and suffered severe reputational harm. The complaint states that officers failed to check Dillon's alibi or confirm his location through phone records and other digital traces.
Legal experts say such cases undermine public trust in policing and highlight the need for stronger safeguards. When an AI system points to a suspect, human investigators must treat it as a lead, not a conclusion. The lawsuit argues that the officers did the opposite: they assumed the machine was correct and then searched for evidence to support that assumption.
Why This Matters
This case illustrates a growing problem as police departments adopt AI tools without clear protocols for their use. A false arrest based on faulty facial recognition can destroy a person's life even if charges are eventually dropped. The legal system provides limited recourse, and many victims lack the resources to sue. Dillon's lawsuit seeks damages for violations of his Fourth Amendment rights against unreasonable seizure and his right to due process.
The outcome could set a precedent for how courts handle claims of AI-related police misconduct. It also puts pressure on state legislatures to require independent verification of algorithmic matches before arrest warrants are issued. For now, the case serves as a reminder that technology is only as reliable as the people who use it. When officers abdicate their investigative duty to a black box, the consequences fall on innocent individuals.
