New Orleans has put automated facial recognition alerts on hold after reporting showed that police had used live camera feeds to identify suspects in real time. The reported system relied on a private camera network and raised questions about secrecy, oversight, and whether arrests were made outside the process required by city rules.
What the camera system was reported to do
A Washington Post investigation found that New Orleans police had spent years scanning live street camera feeds and using facial recognition to identify suspects as they appeared in public. The system reportedly depended on a private network of more than 200 cameras that could send alerts to officers' phones when a possible match for a suspect was detected.
According to the Post, court records and public data suggest the cameras “played a role in dozens of arrests.” But the reporting also found that most uses were not disclosed in police reports, leaving a gap between what the technology may have helped police do and what was formally recorded.
The camera network was operated by Project Nola, a nonprofit founded by former cop, Bryan Lagarde. The Post reported that Lagarde wanted to help police more closely monitor the city’s “crime-heavy areas.”
Why the 2022 ordinance matters
The central issue is not only that facial recognition was used. It is how it was allegedly used.
A 2022 city council ordinance required more oversight around the technology. Under that framework, police were supposed to use facial recognition only to find “specific suspects in their investigations of violent crimes.” The process also required images to go to a “fusion center,” where at least two examiners “trained in identifying faces” using AI software had to agree on a possible match before police approached a suspect.
The Post found a different pattern. It reported that “none” of the arrests “were included in the department’s mandatory reports to the city council.” It also found that at least four people arrested were charged with nonviolent crimes.
That matters because the ordinance was designed to slow the process down before an encounter with police occurred. A live alert sent to an officer’s phone can turn a technical match into an immediate street-level decision. The required review process was meant to create a buffer between an algorithmic suggestion and police action.
The pause and the demands for answers
New Orleans Police Department superintendent Anne Kirkpatrick told the Post that she would review the program and turn off all automated alerts until she is “sure that the use of the app meets all the requirements of the law and policies.”
The American Civil Liberties Union wants a stronger response. The ACLU is calling for a full investigation into how many arrests were made and is urging NOPD to permanently stop using the AI-enhanced feeds.
Alanah Odoms, the executive director of the ACLU of Louisiana, said in a statement sent to Ars that without a full investigation, there would be no way to know the extent of potential harms from secret AI surveillance to the community.
“We cannot ignore the real possibility of this tool being weaponized against marginalized communities, especially immigrants, activists, and others whose only crime is speaking out or challenging government policies,” Odoms said. “These individuals could be added to Project Nola’s watchlist without the public’s knowledge and with no accountability or transparency on the part of the police departments.”
Nathan Freed Wessler, the deputy director of the American Civil Liberties Union (ACLU) Speech, Privacy, and Technology Project, told the Post this is “the first known widespread effort by police in a major US city to use AI to identify people in live camera feeds for the purpose of making immediate arrests.”
Conflicting claims about usefulness and control
Project Nola has claimed in social media posts that the camera network assisted in at least 34 arrests since 2023. The Post said it struggled to verify that claim because “the city does not track such data and the nonprofit does not publish a full accounting of its cases.”
Police records submitted to the city council painted a much narrower picture, saying the network “only proved useful in a single case.” That gap is important because it makes it difficult to assess the actual role facial recognition played, whether suspects were misidentified, and what safeguards were used before arrests.
Lagarde told the Post that police cannot “directly” search for suspects on the camera network or add suspects to the watchlist in real time. Reese Harper, an NOPD spokesperson, told the Post that the department “does not own, rely on, manage, or condone the use by members of the department of any artificial intelligence systems associated with the vast network of Project Nola crime cameras.”
At the same time, a federally mandated 2023 audit said New Orleans police complained that complying with the ordinance took too long and “often” resulted in no matches. That complaint can be read in more than one way. It may suggest a slow process, or it may show that the safeguards were filtering out weak matches before police acted on them.
Why this case reaches beyond New Orleans
The Post noted that New Orleans stood out in the US because the reported practice used live facial recognition camera feeds to support immediate arrests. The Security Industry Association told the Post that four states—Maryland, Montana, Vermont, and Virginia—and 19 cities nationwide “explicitly bar” the practice.
The risks are not theoretical. The Post reported that “at least eight Americans have been wrongfully arrested due to facial recognition.” When a system moves from a possible match to a rapid police response, errors can carry immediate consequences for the person stopped or arrested.
“By adopting this system–in secret, without safeguards, and at tremendous threat to our privacy and security–the City of New Orleans has crossed a thick red line,” Wessler said. “This is the stuff of authoritarian surveillance states and has no place in American policing.”
For now, the automated alerts have been paused while the department reviews the program. The unresolved questions are basic but significant: how often the system was used, how many arrests it influenced, whether the 2022 ordinance was followed, and whether the public can get a full accounting of a surveillance tool that operated largely out of view.