A New Jersey Assembly committee has advanced legislation that would prevent police from relying solely on a facial recognition match to establish probable cause for an arrest warrant. The Assembly Public Safety and Preparedness Committee approved A5423 on Monday in a party-line vote.
Assemblyman Joe Danielsen, a Democrat from Franklin, sponsors the measure. Every Democrat on the committee supported it, while Republican Assemblymen Antwan McClellan of Ocean City and Paul Kanitra of Point Pleasant Beach voted against it. The committee vote advances the bill; the proposed restriction has not become law.
As introduced, the bill defines facial recognition as technology that helps identify a person from an image or video using physical characteristics. It covers state, county and municipal law enforcement agencies and would take effect immediately if enacted. Danielsen’s accompanying statement calls for traditional identification methods to reduce wrongful-arrest risks and says studies have shown facial recognition is less accurate when identifying people of color and women. In June, the New Jersey Supreme Court separately ruled that criminal defendants are entitled to information about the facial recognition technology police used to investigate them.
The ACLU of New Jersey supported the legislation, warning that facial recognition can lead to innocent people being arrested. An attorney for the organization cited a Woodbridge case in which a man spent 10 days in jail despite having an alibi, after police relied improperly on a facial recognition result.
The attorney argued that a false match can influence the rest of an investigation even when police subsequently seek corroboration. Eyewitnesses, family members and parole officers may also identify the wrong person, the attorney said, because the software has already selected someone who closely resembles the suspect.
Kanitra acknowledged the risk of inaccurate results, particularly when investigators use grainy surveillance footage. He nevertheless argued that improvements in artificial intelligence could make facial recognition more reliable than traditional investigative methods and questioned whether the bill anticipated those changes. He called the proposal short-sighted and argued that lawmakers risked depriving police of a valuable investigative tool.
“I think we’re coming to the point where technology may be more accurate than law enforcement’s traditional methods,” Kanitra said. His alternative was to require a minimum confidence threshold for a facial recognition match before allowing it to independently establish probable cause.
“If we had a minimum threshold of a confidence level of what the camera is, I think it would take care of the concerns of everybody,” he said.
Sources: New Jersey Globe, New Jersey Legislature
–
By the ID Tech Editorial Team
