A Fort Myers man is suing the Jacksonville Beach Police Department after he says he was wrongfully arrested because of facial recognition technology that incorrectly identified him as a suspect.

Robert Dillon claims an officer at the Jacksonville Beach Police Department used facial recognition technology to match him with a suspect captured in a surveillance image. That image was taken from the McDonald’s on 3rd Street in Jacksonville Beach where officers were looking for a man accused of attempting to lure a young girl.

But Dillon said he had never been to Jacksonville Beach and the man in the surveillance photo was not him

“I don’t wish this on my worst enemy,” Dillon said.

Dillon was arrested in Lee County eight months after the initial incident.

“I argued with the police officer for 20 minutes and he insisted that I know that I’m looking at your mug shot. It’s just, it’s horrible that this AI is put in the hands of people that aren’t qualified to use it,” Dillon explained.

The charges against Dillon were ultimately dropped. He is now accusing the Jacksonville Beach Police Department of improperly using facial recognition technology.

According to the Pinellas County Sheriff’s Office, which runs the statewide facial recognition database called FACESNXT, the technology is only supposed to be used as an investigative lead and it is not enough to solely be used for probable cause for an arrest.

First Coast News Crime and Safety Expert, Mark Baughman, worked for the Pinellas County Sheriff’s Office. He said they were taught to use it as an investigative tool.

“This is a pointer, you can’t rely solely on this technology to affect an arrest, or at least I don’t think you should. It’s like anything else you’re trying to build the case around it with other factors,” Baughman explained.

The lawsuit alleges that the Jacksonville Beach police officer checked license plate reader records for vehicles associated with Dillon in Duval County around the time of the incident, but found no evidence placing him there. The lawsuit also claims the officer did not follow up with the victim in the incident, but instead had a McDonald’s employee who was working that night but did not see the confrontation pick Dillon’s picture out of a photo lineup.

As questions continue about law enforcement’s use of the technology, experts say the accuracy of facial recognition systems can depend heavily on the quality of the images being analyzed.

Dr. Mengdi Zhu with the University of Florida said newer facial recognition programs use artificial intelligence and biometric data to compare facial features. She said image quality plays a critical role in the reliability of results.

“If the arrest is done based on, let’s say a footage in from the camera, from the security camera, then the quality of that could also be either pretty good or pretty bad and then that would also affect, if you are feeding good amount enough amount of data into the AI, it could be pretty precise, but it could also make mistakes,” Zhu said.

Along with the Jacksonville Beach Police Department, a Jacksonville Sheriff’s Office sergeant and Sheriff T.K. Waters, and the Pinellas County Sheriff are named in the lawsuit.

Both the Jacksonville Beach Police Department and the Jacksonville Sheriff’s Office declined to comment on the case citing pending litigation.

The Pinellas County Sheriff’s Office provided this statement:

“The Pinellas County Sheriff’s Office (PCSO) maintains the FACESNXT facial recognition system. The Jacksonville Beach Police Department (JBPD) is a user on the PCSO system governed by the attached MOU. The MOU requires that PCSO provide training to JBPD officers before they use the system and JBPD is required to ensure that its officers take the training before they are allowed to access the system. That occurred in this case. The assertion in the lawsuit that PCSO failed to train is patently false. Moreover, the training is unequivocally clear that FACESNXT is an investigative tool only. Facial recognition results are never “matches.” Independent investigation is required to determine whether any person in a photo array return in the facial recognition system is the person who committed a crime requires a law enforcement officer to determine probable cause through independent means.

The MOU between PCSO and JBPD requires that JBPD have a policy governing the use of FACESNXT by its officers. JBPD has a policy and it is attached. In section E-5 and 6 of the policy it states that “FACESNEXT is an investigative tool and any law enforcement action taken based on a submission to FACESNXT will be based on the agency’s own identity determination and not solely the results of the FACESNXT results.” “FACESNXT results are strictly investigative leads to assist with subject identification and/or verification.”

Proper safeguards were in place to require that the system be properly used. It was wrong for the JBPD officer to determine probable cause existed to arrest someone solely based on their photo appearing in a facial recognition photo array. Liability for doing that rests solely with any officer who did so, and certainly not PCSO.

Facial recognition is a valuable investigative tool that helps solve crimes and effect public safety. Facial recognition should not be castigated because of human wrongdoing while using the system. This is a people problem, not a facial recognition problem.”