by Sylvia Mendoza
Facial recognition technology (FRT) can be helpful and make life easier, providing security control measures for accessing phones, computers, or smart homes. A broader picture shows how the technology has been integrated into everyday biometric security, allowing entry into public spaces like airports, schools and sports venues.
According to the U.S. Government Accountability Office, an independent, nonpartisan agency that assists the U.S. Congress with investigations, more than two-thirds of police departments across the country use FRT, mostly through body cams, dash cams or phones.
When FRT works, it can be great. Like any technology, however, FRT has flaws. Experts recommend that the technology be used as a tool to help generate leads for police investigations, but not as proof of someone’s guilt. That is because sometimes, the “matches” are incorrect, and innocent people can be misidentified.
“When it comes to criminal prosecution, it is alarming,” says J.D. King, a professor at Rutgers Law School in Camden. “Law enforcement has begun using FRT to identify suspects and charge people with crimes even when little or no other evidence connects them to the crime. Just as with other kinds of forensic technology, like fingerprints, this can be useful but it can also ensnare innocent people in the system.”
Bias baked in
Professor King, whose areas of expertise include criminal procedure, ethics and criminal law, also previously served as a public defender in Washington, D.C. He says two significant issues with FRT are whether the government can and should regulate its use by private (non-governmental) parties, and what safeguards should be in place when the government wants to use this technology in criminal cases.
Studies, including a 2018 Massachusetts Institute of Technology (MIT) report, show FRT systems are less accurate and more biased when identifying people of color and women. For example, the MIT report indicated that there is an error rate of nearly 35% when targeting dark-skinned women.
“This leads to false positives when identifying Black, Latino, and Asian subjects,” explains Professor King. “It compounds and magnifies existing racial bias in the criminal justice system when police departments and FRT companies use databases of previous suspects to train the FRT models. This ‘algorithmic bias’ takes the existing racism of the criminal justice system and makes it worse.”
Poor-quality images can also be inaccurate and produce flawed algorithms. If there is no human oversight in proper investigations, misidentifications can occur. When people are scanned without their permission or knowledge that also creates privacy issues.
According to the American Civil Liberties Union, the Federal Bureau of Investigation’s FRT database includes hundreds of millions of photos. Without legal guardrails for these databases, invasive surveillance can occur and violate people’s rights.
“Technology now allows for a level of scrutiny and surveillance that is unlike anything we could have imagined in the past,” explains Professor King. “Many European countries, and some U.S. states, are rethinking whether modern society needs a broader definition of privacy. With license-plate reading cameras, facial recognition technology, and linked security cameras, governments and private companies can track intimate details of our lives.”
A New Jersey case
The 2023 case of State v. Arteaga changed the rules for criminal defendants in New Jersey that are identified by FRT. The case involved Francisco Arteaga who was identified as a potential suspect seen on surveillance footage holding up a store in West New York, NJ in 2019. New Jersey police officers initially sent images from the store’s cameras to the New Jersey Regional Operations Intelligence Center, where no matches were found. The detectives assigned to the case then sent the raw footage to the New York Police Department’s Real Time Crime Center database and identified Arteaga, who was living in Queens at the time, as a possible match. He was indicted even though he had an alibi.
“The Arteaga case had to do with what rights a defendant has when the government has used FRT as evidence against them in a criminal case,” Professor King says. “When the store was robbed and police responded, the victim could only describe the robber as a ‘Hispanic male wearing a black skully hat.’”
When prosecutors refused to disclose information about how the FRT system that identified him worked, including the probability for error or bias, Arteaga’s attorney filed a motion to disclose the information. The trial judge in the case denied the motion. Arteaga appealed and a three-judge panel of the New Jersey Superior Court, Appellate Division sided with him.
“Here, the items sought by the defense have a direct link to testing FRT’s reliability and bear on defendant’s guilt or innocence. Given FRT’s novelty, no one, including us, can reasonably conclude without the discovery whether the evidence is exculpatory or ‘merely potentially useful evidence,’” Judge Hany Mawla wrote for the three-judge panel. “Defendant must have the tools to impeach the State’s case and sow reasonable doubt.”
Professor King explains that the appellate court found that the failure to share information about how the system can misidentify suspects deprived Arteaga of his constitutional rights. In other words, Arteaga has the right to challenge the evidence presented by law enforcement.
“Although the court’s ruling only applies in the context of criminal prosecutions, it shows that at least some courts are concerned about the potential for abuse and misuse of this new technology and are looking to ensure that evidence used in criminal cases is reliable,” says Professor King.
“What this case really warned us about is the threat of unchecked surveillance power and the acquisition of all of this surveillance technology with no accompanying accountability framework,” Dillon Reisman, a staff attorney with the ACLU-NJ, told The New Jersey Monitor. “All of these systems, by their nature, involve interstate cooperation and systems that are bigger than one individual agency, and that makes it extremely difficult to have any sort of transparency, to learn how the system is used, to learn how the system might be flawed, and to advocate against it.”
States ban FRT
In 2023, Alexandra Reeve Givens, President & CEO of the Center for Democracy & Technology, testified before the U.S. Senate Judiciary Committee’s Subcommittee on Human Rights and the Law.
“Given the range of risks facial recognition poses to civil rights and civil liberties, there is not a silver bullet policy solution: lawmakers need to enact a broad set of safeguards to prevent harm both from misidentifications and misuse,” Givens testified.
Currently there is no federal law banning FRT or restricting law enforcement’s use of it. However, on the state level, more than 20 states have enacted or expanded existing laws to restrict the mass scraping of biometric data or limit police use of FRT, according to the National Conference of State Legislatures, For example, New Hampshire and Oregon ban the use of facial recognition and police body cameras.
The Illinois Biometric Information Privacy Act regulates how biometric identifiers, such as fingerprints and facial geometry, are handled by private entities and makes it illegal to collect, use, or store biometric data (including face scans) without written consent. A Maryland law limits police use of FRT to the investigation of serious crimes, including human trafficking, violent crimes or hate crimes. The law also requires prosecutors to notify defendants that facial recognition was used in an investigation, leading to the charges brought against them.
The New York Police Department instituted a policy regarding facial recognition in 2023. The policy says, “FRT can be used to identify potential persons of interest but requires corroborating evidence before action is taken. The facial recognition process does not by itself establish a basis for a stop, probable cause to arrest, or to obtain a search warrant.”
“Facial recognition technology has so many beneficial uses but also can be abused and misused so easily,” says Professor King. “We are only beginning to grapple with how much we want to regulate or allow this technology.”
Discussion Questions
- Does the loss of privacy as described in the article concern you? Why or why not?
- What do you see as the benefits of facial recognition technology? What are the drawbacks?
- If you were to craft a law limiting the use of facial recognition technology, what would you include?
Glossary Words
algorithm—a set of rules to be followed in calculations.
appeal—a request that a higher court review the decision of a lower court.
biometric— involving the automated recognition of individuals by means of unique physical characteristics, typically for the purposes of security.
exculpatory—to clear of blame; evidence that is exculpatory would clear a defendant from guilt.
indicted—to be charged with a criminal act by a grand jury.
motion—a proposal made to a court.
nonpartisan—not adhering to any established political group or party.
reasonable doubt—the high legal standard set in criminal cases that requires jurors to acquit [free] a defendant if they have a substantial, logical uncertainty about guilt based on evidence or lack thereof.
This article originally appeared in the spring 2026 edition of Respect, NJSBF’s diversity and inclusion newsletter.