Security robots are starting to join the ranks of human guards. Here’s what we know

From New York to Hawaii, people can spot security robots equipped with a suite of sensors patrolling the perimeters of some residential communities and apartment buildings.
Security robots are rapidly becoming a viable complement to human security guards, sparking concerns about privacy issues and potential job losses in the security field. Critics are also raising questions about their effectiveness of keeping the public safe.
Showing their capability as a reliable security solution is tough – there is lack of public data to prove it. However, experts and manufacturers say their true strength lies in working alongside security officials with their advanced technologies.
Here’s what we know about security robots:
Security robots possess an array of tools that humans lack, making them unique partners to security systems:
Security robots can operate 24/7 and excel in the repetitive aspects of the job, like sitting at a post or walking a set route. Robots allow humans to perform hands-on tasks that require empathy and sympathy.
The K5 security robot is the most well-known product of California-based robotics and security technology company Knightscope.
These robots are not equipped with any form of deadly force, according to Knightscope co-founder and Executive Vice President Stacy Stephens.
Knightscope is not the sole manufacturer of security robots either. AI and robotics company Cobalt AI manufactures a security robot that patrols hallways, office spaces and indoor facilities.
It has a built-in screen, which enables real-time human-to-human interaction, allowing remote communication between security officials and people on-site.
Advanced robotics company Boston Dynamics manufactures a line of robots called “Spot,” which is used by police departments, manufacturers and construction companies. Its primary objectives include detecting dangerous gases like carbon monoxide and navigating hazardous environments unsafe for humans.
Jay Stanley, a senior policy analyst for the American Civil Liberties Union, acknowledged the unique capabilities of security robots lie in completing tasks and navigating areas that are unsafe for humans.
New York Mayor Eric Adams speaks during a press conference at the Times Square subway station about a Knightscope K5 autonomous security robot that was set to patrol the station. (Barry Williams / NY Daily News / Getty Images via CNN Newsource)
Security robots are roaming some streets in Atlanta and San Diego, among other U.S. cities, and their presence is expanding into police departments, where they are being integrated into law enforcement operations.
In 2023, New York Mayor Eric Adams and the New York City Police Department unveiled a Knightscope K5 to patrol Times Square and the city’s metro system while accompanied with a police officer.
This February, NYPD confirmed the robot had completed its trial run and was subsequently removed from the streets. No additional information was available about why the robot did not continue serving the city.
Earlier this year, Massachusetts State Police deployed a Boston Dynamics Spot, a four-legged robot, during a seven-hour standoff with a suspect in Boston, according to CNN affiliate WHDH. Roscoe the robot took a bullet while trying to locate the gunman.
Three K5 robots were deployed in San Diego earlier this year, according to CNN affiliate KFMB. One is monitoring the area of an apartment community in Claremont, where it operates 24/7 to keep residents safe and ward off car thieves, according to the robot’s operator.
Earlier this month, an apartment building in Atlanta deployed a K5 security robot, according to CNN affiliate WANF. It roams the sidewalks outside the buildings to monitor the perimeter and protect residents.
John Hassard, a loss prevention and security expert with Robson Forensic, says the greatest strength of security robots is their ability to serve as a highly customizable extension to an existing security system.
“One would assume these are not entry level products, so if someone buys these, they already have a pretty good camera system that they’ve optimized reasonably,” he said. “This extends that. It makes that camera system more valuable.”
For example, Knightscope’s robots are designed to collaborate with existing security infrastructure and surveillance systems, according to Stephens.
Knightscope provides security software, which allows the robots to issue alerts when an anomaly is detected. Depending on the situation and the sensor used to detect an issue, an alert will be broadcast to the security system or department where the robot is deployed and ask for a guard to investigate an issue.
Robots can also be programmed for various outputs based on the sensors they wield. Robots deployed in a parking garage with license plate recognition can create lists of license plates to flag and alert to security staff.
Hassard also believes security operations can cut down on the number of guards they deploy with these devices as long as a location already has effective surveillance infrastructure in place.
“You could cut down the number of security officers working, replacing them with this,” he said. “By default, this thing doesn’t take breaks, doesn’t fall asleep, and you absolutely know what its responses are going to be.”
Experts and one robot maker CNN spoke to agree on the deterrence capability of security robots.
“When people come onto a campus and you see this large, 5-and-a-half-foot tall, 3-foot-wide, 400-pound robot that says security or police on it, that’s what people start to think about,” Knightscope’s Stephens said.
The robots’ ability to serve as a physical deterrent can help defuse situations that may escalate if a police officer were present instead, according to Paul Scharre, executive vice president at the Center for a New American Security.
“If someone vandalizes a robot and bashes it up, you’ve got a video recording of them committing an act of vandalism, and you find that person and prosecute them,” he said. “No lives are lost, no people are harmed in the incident.”
However, Scharre explained the robots can provoke incidents, as they have a physical presence.
“If the robot is sort of labeled as a security robot or police robot, it could be perceived by people as being intrusive and interfering on their privacy,” he said.
What’s not known, however, is what the data says about how effective these robots are in deterrence, Hassard said. The lack of data could stem from companies not wanting to admit they had security issues in the first place, he added.
“Deterrence is a huge thing in security because we don’t want to catch people doing stuff. We want to keep them from doing it, which is tough to measure,” he said.
Knightscope, founded in 2013, deployed its first security robots in the US in 2016. (Knightscope, Inc. via CNN Newsource)
Stanley from the ACLU said security robots can be a privacy nightmare for everyday citizens.
“If these robots are making decisions about who to watch based on some AI, that raises enormous questions about profiling, fairness and transparency,” he said.
The lack of transparency of the robots’ algorithms and intelligence is what worries Stanley. Should someone have an adverse interaction with the robot, there should be legal liability protections set in place by whoever is operating the robot, he said.
“And that’s true with any sort of AI device, whether it’s a robot or just an algorithm running in software,” he said.
For example, Knightscope’s robots do not have access to national criminal databases and their mobile models are not equipped with facial recognition software, according to Stephens. However, Knightscope manufactures a stationary model of the K5 that can use facial recognition software.
People need to start thinking about how they should interact with robots as they become more advanced, according to Scharre.
“Things to think about as you start deploying security robots, are all about how we interact with the technology, how do we perceive it, how do people respond to it?” he said.
A future with more advanced and intelligent security robots capable of automated enforcement and more automated surveillance could be a problem that would need to be addressed in the future, Stanley said.
While the vision of security robots patrolling our streets feels like a glimpse of what’s to come, that future remains uncertain without a proven track record to back it up.
These robots are not immune to malfunction, nor are they flawlessly programmed. In 2016, a K5 knocked down and slightly injured a toddler in a California mall. The next year, another K5 plunged itself into a fountain in a Washington, DC, office building.
Those malfunctions and the lack of data beg another question. Stanley questioned why any security or police department would choose to purchase a robot over traditional options like static surveillance systems or human security guards.
“It’s hard for me to imagine that they’re going to work out anytime soon in the marketplace when there are other technologies that can do a job, and also when human beings can just do the job.”