- AI-enhanced camera networks are increasingly used for public safety in Utah.
- Concerns arise over privacy and ethical use of surveillance technology in society.
- Experts stress balancing technology benefits with privacy rights and transparent practices.
SALT LAKE CITY — Surveillance cameras have stood for decades as symbols of safety and security from homes to businesses to cities.
Today, though, those cameras have increasingly transformed into camera networks enhanced by artificial intelligence, and they're used to recognize everything from license plates to lost dogs.
Police say that technological capability can be a great thing when it comes to public safety.
"Leveraging the technology that we have right now can quickly take a public safety concern and locate that individual much faster," said Tanner Jensen, chief of investigations with the Utah Department of Public Safety.
However, what some have characterized as a "surveillance state" is increasingly being weighed against the cost to personal privacy.
"You go out in public, and you go into a public place, you don't have any privacy," Debra Young said as she headed down Salt Lake City's Main Street. "Who doesn't have cameras nowadays? Everybody has cameras."
It was a similar perspective in Provo's downtown from Jennifer Harlow, who had taken note of the high volume of doorbell cameras around her.
"Even with our consent, it's like, 'Oh, am I consenting for everyone in my neighborhood? " Harlow questioned.
From home surveillance cameras to professional-grade cameras mounted on buildings and in intersections to smartphone cameras, privacy advocates and industry professionals acknowledge it's hard to go anywhere and not wind up on video.
"On one hand, it's a 'surveillance state' issue — like, are we being watched too much?" David Notowitz, founder of the National Center for Audio & Video Forensics, said. "We have a right to privacy in the United States, and I hope that we can maintain it in some way."
Solving crimes
Numerous police agencies in recent years have turned to automatic license plate recognition cameras during investigations to identify suspect vehicles and solve crimes.
Those cases have frequently made headlines, including in March, when automatic license plate recognition cameras aided in the arrest of a stranger from Iowa suspected in three random killings in Wayne County.
Also recently, police used the technology in the Ogden area to spot a stolen SUV and eventually locate a homicide suspect from Nevada.
Provo Police Capt. Robert Patrick said the department's network of over 20 Flock Safety automatic license plate recognition cameras has already helped to solve "dozens and dozens" of cases, including serious crimes like kidnapping and automobile homicide, since the system was implemented in the fall of 2023.
"I can say it's a very exciting thing to be able to go back a period of time, because crimes aren't always reported the moment that they happen," Patrick said. "To be able to go back and to have potential evidence that's able to be collected and gleaned that leads to a suspect and an apprehension, it's a good thing. It can be a good thing."
Patrick said the system can only be accessed for investigative reasons and explained that it can't be used to track people across the city's 640 miles of roadway.
"If I only have 21 data points and 640 miles, there's not a lot of tracking I can do, but there is some general information I can get about whether a vehicle was in a location at a specific time that could be helpful in a criminal investigation," said Patrick as he explained how the system works. "Think of a trail camera. A trail camera is a camera that's posted on a tree, or in this case, it's a camera that's mounted on a pole, and anytime a vehicle with regard to an ALPR camera passes that camera, it takes a series of photographs. Depending on how fast the vehicle is going, it may capture one photograph or two or three, and those photographs are then time-stamped, dated and attached to those particular photographs. Those photographs are then transmitted to a secure site, where they're stored for a number of days, and they're searchable in that storage database."
Patrick cited a recent 30-day period in which the system captured a rough total of 540,000 images of various vehicles.
"That's a lot of photographs and a lot of images that they do capture," Patrick acknowledged.
He said not only are license plates searchable information, but so are specific descriptions of rear-end damage and things like window stickers or other items that may be visible around the back of a car.
Other camera networks and AI
License plate reading camera networks are not the only kind to leverage recognition technology and artificial intelligence.
Ring made headlines earlier this year with its Super Bowl ad about its new "Search Party" feature, which utilizes its customers' doorbell cameras to locate lost dogs.
The company rolled out its "Fire Watch" feature, which operates in partnership with nonprofit Watch Duty to notify Ring customers whenever a nearby wildfire is identified.
According to information from Ring, cameras from customers subscribed to Ring Protect within an alert zone will start to analyze video for "visual signs" of smoke or flames and signal an alert to customers in the area.
Surveillance concerns
Notowitz, whose National Center for Audio & Video Forensics firm is retained by defense attorneys and police alike to recover, analyze and clarify video and audio evidence, said he has watched the progression of AI systems in both home surveillance and license plate reader networks.
He said that while recognition technology in doorbell cameras has not done a great job yet in recognizing faces, it is possible for those cameras to be programmed that way as well.
"You can do searches on your own materials now," Notowitz explained. "They'll recognize faces for you. You can actually instruct it and say, 'I have five faces I want you to keep track of for me, so if they show up, send me an alert.' There are things like that that they'll do for you as a service, so let's say, God forbid, you have someone that you're nervous about, you can program those into your cameras and keep you up-to-date if those faces show up."
Notowitz said artificial intelligence has greatly enhanced the ability to make databases from surveillance camera networks searchable.
"The AI tool now can do things that police can't do and that computers couldn't do previously, which is develop, kind of, systems of logical tracking that a human being doesn't have the time to figure out," Notowitz said.
He said police capabilities are often well beyond what they'll disclose or acknowledge publicly, including with automatic license plate recognition networks.
While police have maintained the networks aren't good for tracking a direction of where a vehicle is going, Notowitz said with the proper connections, those camera networks can be utilized to map every place a car has been — even across the country — potentially for a period of months.
"It's not just this car," Notowitz explained. "As you'll see in cases, they'll find connections between this car and another car, and they'll say, 'Who has come in contact with this main car more than five times in the last two months?' They can do that, and they'll find other license plates that are parked near, or that drive near, the other license plate."
Notowitz said it isn't just police that use the technology, and he used shopping malls as an example of a type of business that has been known to sell that kind of information to third parties.
"In malls, they have license plate readers," Notowitz said. "You go and do parking, and they have license plate readers. There are license plate readers in a lot of places you don't think about. Like, I said, in malls, did you even realize that they're tracking you there quite a bit in your car? They do it for civil reasons. Like, when you get a loan on your car and you don't pay off the bill for three months, they'll start maybe tracking you and malls do that. They sell the data to private companies — private companies get this data. Police do too, if they pay for it and they have deals with people, but they also have a lot of government cameras that get data all the time."
Notowitz acknowledged the benefit of police being able to identify and apprehend suspects more quickly, especially when it comes to violent crimes, but he also said that level of surveillance has concerned him.
"It's actually very disturbing to me that the level of surveillance now going on with the license plate readers and the level of investigations that are going on," Notowitz said. "How much is my license plate being analyzed? They don't have the right to do that. I didn't do anything wrong. I'm not on any suspect list, but they're doing it."
Former Salt Lake City Police Chief Chris Burbank also expressed reservations about AI-enhanced camera networks.
"That's tremendous technology, and you look at that, and you say, in that circumstance, that's fantastic — that's exactly what we want, and we really need it — but what is the cost to that for everyone else that is not involved in criminal activity.
Burbank noted the extra hundreds of thousands of images that license plate reading networks capture of cars of innocent people.
"Is the one good time worth the how many, thousand, how many million times that it's not important," he continued. "Are we getting better or are we sacrificing our freedoms and well-being in order to have that one-in-a-lifetime chance?"
Burbank said he was also concerned with how long the data was being retained, noting it can vary across municipalities and departments.
"I see the value from my former profession of having more information, and I know exactly what some of the capabilities are, but, boy, it comes with tremendous responsibility, and are we actually protecting the well-being of everybody as opposed to catching the one time," Burbank said. "I'm always a firm believer that public safety is done best in preventing crime from occurring, not chasing it after the fact, and we seem to be putting a lot of eggs in the basket of after the fact as opposed to let's talk about how we improve society on the front end so we don't have the consequence to deal with."
Ethical use
Patrick underscored that the ethical use of license plate reading cameras and other camera networks is important to Provo police, and noted the department only retains license plate reading images for 30 days, and after that, the images are no longer searchable.
Patrick said transparency was important with current and future uses of the technology, and he said the department's automatic license plate recognition system comes with a transparency portal, as well as regular auditing of investigations to ensure the cameras were not improperly accessed.
"I don't ever want the public to look at this program and say, well, it was used improperly, and we didn't know about it," Patrick said.
He said cameras can also be an important tool in exonerating people.
"People get descriptions of vehicles wrong or descriptions of suspects wrong, so to be able to verify that through an actual physical photograph of a potential suspect vehicle is quite helpful to either place the person at the scene of the crime or place them somewhere else."
At the state level, Jensen said ethical use was also important at the Utah Department of Public Safety.
"Law enforcement and, I think, government as a whole in our society is concerned about the use of technology and artificial intelligence and how it may impact privacy rights, and so I think there's a lot of things here in the State of Utah that we're doing to make sure we can balance that use of technology, use of artificial intelligence and maintain civil rights, privacy rights," Jensen said.
He said police are working with lawmakers to establish guardrails around when and how the technology is used, as well as transparency measures and auditing capabilities, and he believes that progress has already been made on those fronts.
"I think the technology is growing quickly," Jensen said. "I think they've created a really good foundation on how to leverage technology."
Ring, meanwhile, issued a statement to KSL about the use of artificial intelligence capabilities in its services to customers.
"Ring's mission has always been to make neighborhoods safer, and we recognize that new technology can raise important questions about privacy," said the statement from a Ring spokesperson. "Our AI-powered features are designed for specific use cases — like helping reunite a lost pet or identify a nearby fire — and are built to give customers meaningful context about what's happening around them and the ability to help their neighbors if they choose. Customers can enable or disable these features and always choose whether and when to share their video through them. We've designed these features to be useful while keeping privacy and customer control central, and this approach continues to guide how we develop new technology."
Harlow, a mom concerned about the world where her children are being raised, said artificial intelligence and camera networks can be a positive in society.
"I think AI is a great thing, but I think we should be a little more careful with how we're using it," Harlow said. "It's like, where is that data going? It's kind of scary."
