Flock Faces Backlash, But AI Crime Centers Draw Scrutiny

As the police surveillance company Flock Safety confronts a wave of vandalism, contract cancellations and congressional scrutiny, privacy experts argue that the public debate is focused on the wrong target: the far more powerful AI-powered tools that fuse data from across a city into a single surveillance dashboard.

Andrew Guthrie Ferguson, a law professor at George Washington University, told NPR that he has watched the controversy over Flock’s automated license plate readers (ALPRs) “with a little bit of puzzlement.” The bigger concern, he said, is the software powering so-called real-time crime centers, which combine not just license plate cameras but video analytics, drones, body cameras and more into “a visual single-pane-of-glass vision of what’s going on in a city, with time machine capabilities where you can go back and investigate crime.”

A Backlash Against Flock

Flock Safety has become a lightning rod in recent months. NPR has documented the company’s cameras being vandalized in at least 36 states, and at least 100 cities have switched their cameras off or canceled contracts with Flock or similar vendors, according to an advocacy group’s count. Several cities have acted after discovering that their local data had been shared more broadly than officials knew. As NPR reported, many city officials learned only after the fact that the data was shared more widely than they had intended — sometimes reaching federal agencies. Flock says customers control their own sharing settings and that it has strengthened oversight and audit capabilities, but the revelations eroded trust in communities such as Flagstaff, Arizona, and Santa Cruz, California. Santa Cruz council member Susie O’Hara told NPR she was “very dissatisfied with a multibillion dollar company continuing to make mistakes and putting our local data at risk.”

The Bigger Picture: Data Fusion

Yet experts say that while ALPRs like Flock’s are visible and easy to rally against, the underlying architecture of police surveillance is expanding faster and with less scrutiny. That architecture is the modern real-time crime center — a central hub where analysts monitor feeds from across a jurisdiction — increasingly powered by AI data-fusion software. The market leader, according to researchers and police agencies, is Fusus, owned by the public-safety technology giant Axon.

Axon describes Fusus as a “rapidly deployable real-time crime center in the cloud.” It functions like the brain of a crime center, synthesizing data from body cameras, dashcams, 911 dispatch, gunshot detectors, surveillance cameras, drones, and sometimes private business cameras that opt in. In 2024, the technology news site 404 Media reported that just over 100 law enforcement agencies had deployed Fusus. According to Axon, that number has more than doubled since.

Police agencies that use the platform praise it as a force multiplier. “We could put analysts in a room and have them access 40 different systems,” Benjamin Weir, a captain with the Oklahoma City Police Department, told NPR. “But that slows them down. So the biggest leveraging factor for a real-time information center to be productive is to have software that can incorporate all these things into one usable area.”

Quincy Whitner, a lieutenant with the Greenville County Sheriff’s Office in South Carolina, said Fusus helped his department find a man who shot a police officer and fled, by combining city surveillance footage with license plate data. “Instead of having to go to a million different systems, a million different places to find answers, we can get it in one platform in one place somewhat instantaneously,” he said.

The Concern: Removing ‘Friction’

Privacy advocates do not dispute the tool’s utility but warn that its convenience is precisely the problem. Rachel Levinson-Waldman, director of the Liberty and National Security Program at New York University’s Brennan Center for Justice, said that even a few decades ago police faced more hurdles before surveilling someone — physically following a suspect or conducting a stakeout took time, staffing and resources.

“In addition to all of the things that you would want in place requiring some kind of criminal predicate to launch it, there is just this kind of organic friction that imposes some constraints,” she said. “As that friction gets lifted, it becomes easier and easier to do investigations, to look at people and even to pull up broad-scale information.”

A Brennan Center report published in November 2025, co-authored by Levinson-Waldman, warns that AI-driven data-fusion tools can perpetuate bias, produce inaccurate conclusions, and enable indiscriminate surveillance. The report found that systems drawing on biased historical inputs risk generating skewed outputs, and that tools relying on natural language processing have misclassified words and struggled with irony and context — including a case in which an AI system flagged a British tourist’s slang as a threat.

Police departments defend the technology. “I understand people’s concerns, but if you’re not up to no good, we’re not looking at you,” Michelle Iezzi, deputy director of the Operational Intelligence Division at the Montgomery County Police Department in Maryland, told NPR. “We are looking at people who are committing crimes.”

Axon did not respond to multiple requests for comment from NPR. But CEO Rick Smith said on the company’s latest earnings call that it values privacy: “We just have to be thoughtful and understand those concerns and say, ‘Look, this is a tool to amplify human ability, not to replace human judgment.’”

What’s Next

Courts are moving slowly in deciding when such surveillance crosses a constitutional line. Ferguson, the George Washington law professor, argues that lawmakers could act sooner by limiting how long data is stored, mandating regular audits, and restricting surveillance near sensitive sites such as abortion clinics or immigration courts. He said those conversations should happen before a tool is ever deployed.

The stakes cut both ways: surveillance technology can sweep up innocent people by mistake, while police say it helps solve crimes quickly. As cities continue to debate their contracts with Flock and other vendors, experts warn that the quieter, more pervasive expansion of AI-powered crime centers may define the next chapter of the surveillance debate — whether or not the public is watching.

“What happens if there is a government that wants to silence a particular group or even a particular person?” Ferguson asked. “And you just built the architecture of surveillance that makes that super-easy.”