Police technology is moving faster than many rules written to contain it. A new AI tracking method described by MIT Technology Review shows how agencies can follow people through attributes such as body size, gender, hair color and style, clothing, and accessories, rather than through facial recognition.
That distinction matters. Local laws limiting police use of facial recognition have been passed around the country, but a system that does not technically use biometric data can still raise similar concerns about surveillance, accountability, and public control.
Why this AI tracking tool matters
The tool at the center of the report offers a way to search for and track people without identifying them by face. Instead, it relies on visible characteristics: body size, gender, hair color and style, clothing, and accessories.
Advocates from the ACLU, after learning of the tool through MIT Technology Review, said it was the first instance they had seen of such a tracking system used at scale in the US. They also said it has a high potential for abuse by federal agencies.
The concern is not only technical. It is also political and civic. The source article notes that the Trump administration is pushing for more monitoring of protesters, immigrants, and students, which makes the prospect of stronger AI surveillance especially alarming to advocates.
Facial recognition bans are often framed around the dangers of matching a person’s face to an identity. But the new tracking method points to a broader issue: police may not need a face to follow someone across video or other sensor data. If a tool can locate a person by a combination of visible traits, the practical effect may still feel like being tracked.
Police AI is expanding quickly
The report places this tool inside a larger shift in police technology. Six months earlier, the author attended the largest gathering of chiefs of police in the US and found departments already using AI for tasks such as writing police reports.
Since then, the technology has continued to develop. Police departments in the US have extraordinary independence, according to the article. There are more than 18,000 departments in the country, and they generally have wide discretion over what technology they buy with their budgets.
That independence gives local agencies room to experiment. It also means there is no single, consistent public process that governs how every department evaluates AI surveillance tools before deployment.
Companies including Flock and Axon sell suites of sensors such as cameras, license plate readers, gunshot detectors, drones, and AI tools that help interpret the data those systems collect. Departments say these technologies can save time, ease officer shortages, and reduce response times.
Those goals are easy to understand. Faster response and less administrative work are practical benefits for agencies under pressure. But the same systems that promise efficiency can also create large pools of information about public life, and AI can make that information easier to search, connect, and act on.
The line between efficiency and surveillance
The central question is whether AI police tools are being adopted with enough transparency and oversight. A department may describe a technology as a way to work faster, but the public may experience it as a system that watches, classifies, and follows people more easily than before.
That tension has already appeared in Chula Vista, California. The police there were the first in the country to receive special waivers from the Federal Aviation Administration to fly drones farther than normal. The department said the drones would help solve crimes and get people help sooner in emergencies, and the source article says they have had some successes.
But the same program has drawn criticism. A local media outlet sued the department, alleging it had reneged on its promise to make drone footage public. Residents have said drones overhead feel like an invasion of privacy.
An investigation also found that the drones were deployed more often in poor neighborhoods, and for minor issues like loud music. That example shows how a technology introduced for emergency response can become controversial when people question where it is used, why it is used, and whether the public can inspect the results.
AI tracking software raises similar questions. If a system can search for people using clothing, hair, accessories, and body characteristics, communities may want to know when it is used, what data it searches, who can access it, and how long records are kept. The source article does not provide those deployment details for the Veritone tool, because Veritone said it could not name or connect the reporter with departments using it.
Facial recognition rules may not be enough
Local laws against police use of facial recognition show that some communities have taken a firm position on biometric surveillance. But the new tracking method suggests that rules focused only on faces may leave room for workarounds.
Jay Stanley, a senior policy analyst at the ACLU, said there is no overarching federal law that governs how local police departments adopt technologies like the tracking software discussed in the report. Departments usually have the leeway to try a tool first, then see how communities respond after the fact.
That sequence puts the burden on the public to react once a system may already be in use. It also gives vendors and departments a chance to frame a tool narrowly, even if its effects are broad.
Stanley argued that the tracking software poses many of the same issues as facial recognition while avoiding scrutiny because it does not technically use biometric data. His warning was direct: “The community should be very skeptical of this kind of tech and, at a minimum, ask a lot of questions.”
He also described steps police departments should take before adopting AI technologies. Those steps include public hearings, community permission, and clear promises about how systems will and will not be used. He added that companies making these systems should allow independent testing.
What communities should ask before adoption
The report does not say that every use of AI in policing is the same. It does show that the rules around police AI are lagging behind the tools themselves. When departments can buy and deploy powerful systems with broad discretion, public oversight becomes a central issue.
Before agencies adopt AI tracking tools, communities can reasonably press for answers on several points:
- What problem is the technology meant to solve?
- What data will the system search or analyze?
- Who can use the tool, and under what conditions?
- What uses are prohibited?
- Will independent parties be allowed to test the system?
- How will the public know whether promises are being kept?
These questions matter because AI can turn scattered sensor data into something more searchable and actionable. Cameras, license plate readers, gunshot detectors, drones, and other systems already collect information. AI tools can make that information easier for police to process at scale.
Stanley’s final question captures the policy challenge: “Are these powers we want the police—the authorities that serve us—to have, and if so, under what conditions?”
That is the real issue behind AI police tracking. The technology is not only about whether a face is scanned. It is about whether communities have a meaningful say before surveillance capabilities become routine.