The technology analyzes footage to detect activity and determine a best course of action. This can include directly speaking to individuals with personalized, AI-generated voice warnings, without human intervention.
Steve Lindsey, LVT's chief information and technology officer and co-founder, explained to Government Technology that major law enforcement agencies in San Francisco and Detroit use LVT mobile surveillance technology for real-time event awareness.
The mobile surveillance units include three cameras mounted on a 22-foot mast, floodlights, strobes and speakers for broadcasting warnings.
Now, Lindsey says, LVT’s surveillance capabilities have been enhanced with agentic AI that he asserts will allow officers to focus their efforts on “really important issues a human has to be involved with.”
LiveView Technologies provided Government Technology with the example video below.
In the video, two individuals approach the side of a parking ramp with a blanket. Keywords appear on the screen describing what the individuals in the shot are wearing and holding.
Soon after their arrival, an automated voice delivers a message: “Attention, the individual in the brown sweatshirt, and the individual wearing the black beanie near the parking garage entrance, please leave the premises immediately.”
Lindsey explained the newest update with the technology uses contextual detection as well as generative AI behavioral deterrence. He said the new tech doesn't just automate tasks; it gives AI agents the ability to make smart decisions based on evolving situations — such how to react to different scenarios.
“As the camera is watching the parking lot, and it’s seeing an active shooter situation versus a loitering situation, it has to make those decisions on, 'Am I going to escalate this to the human?' or 'Am I going to try to use my deterrence capabilities to change behavior?' And that’s where that intelligence comes into play, it knows what should escalate immediately to humans and what it shouldn’t,” said Lindsey.
While in the past the surveillance company used rules-based systems, the updated process also allows the technology to learn from new data, recognize patterns, adapt responses and predict potential threats before they escalate.
“That’s where this automated deterrence comes into play,” said Lindsey. “So we are able to take strobe lights and spotlights as well as audio talk to be able to scare those people away or change behavior.”
LiveView Technologies
“Any of these areas that are along those lines, we’re seeing immediate benefits,” he said. “Officers can focus on other things, then we’re getting better coverage in the other areas.”
However, some digital privacy researchers are skeptical of the true potential of such talking AI crime prevention systems.
“With a lot of these use cases, if you aren’t bothered by the camera in the first place, are you bothered by a robot voice yelling at you?” questioned Dave Maass, investigations director for the nonprofit Electronic Frontier Foundation. “The thing that’s really worrisome is AI making decisions based on what it thinks people are doing. It’s very, very primed for failure and making mistakes. How can an algorithm suss out the dynamics of human behavior?”
However, Lindsey asserts that the cameras alone aren’t being used to prosecute crime, rather, prevent it.
While he does see potential for law enforcement to use agentic AI capabilities within the camera systems to perform rapid video searches to build cases and investigations for prosecution, he noted that in those cases, humans would continue to play a critical role.
“A human is still the one who's looking at video evidence and making a final decision,” said Lindsey.
