ZeroEyes was founded in the aftermath of the 2018 shooting at Marjory Stoneman Douglas High School in Parkland, FL—the deadliest mass shooting at a high school in U.S. history. Today, the company is transforming security in schools, banks, and the cannabis industry through real-time AI surveillance that integrates with existing security systems. I spoke with Rob Carter about the origins of ZeroEyes, and how high-tech weapons detection can mitigate gun violence.
Mary Jane Gibson: How did ZeroEyes get its start?
Rob Carter: The daughter of our CEO came home from middle school one day in 2018 after an active shooter training and asked her dad, ”Is this going to happen in my school?” He and his co-founders [who were all ex-military] got together with the idea that something needed to be done. They walked around schools looking at security cameras, and asked superintendents and police officers what they were used for. The answer was that, after a fight, officials would go back to the security cameras to see what had happened.
The guys in the group knew that object recognition through security cameras was possible, and they wanted to know, “How do we detect a gun?” So they all quit their corporate jobs, moved into our CEO’s basement, and started building ZeroEyes from scratch.
MJG: How did they land on using AI in the surveillance space?
RC: One of the co-founders, our chief product officer, had some experience [in the sector], and knew folks working in computer vision, which is a subset of AI that includes facial recognition and license plate readers. The founders of ZeroEyes knew that figuring out how to detect a gun would be a challenge, but they got their start based on existing object-recognition computer vision models.
MJG: After the research and development period, where was ZeroEyes first deployed?
RC: There were a lot of tests; we ran them at local school districts on weekends and nights, and at a giant water park on the East Coast during the off-season. We got our start in a school district in New Jersey that was willing to be the first to let us go in to test and train the system. We built the algorithm out from there, and we’ve grown to 44 states since 2018.
MJG: When did ZeroEyes enter the cannabis space?
RC: We were pretty hyper-focused on schools, trying to deliver a solution to a problem that has seen an exponential rise since 1999, but we started to see demand from commercial businesses, and we realized that it wasn’t just a school problem; there’s gun violence throughout [the country]. We saw that, whether it was a bank, dispensary, or someone’s private office, there were potential threats happening, and people wanting to protect themselves, their businesses and employees. We started working with cannabis businesses around 2020 or 2021, during the return to work from COVID-19.
MJG: Which is when cannabis businesses were deemed essential.
RC: Yes.
MGJ: How does ZeroEyes balance surveillance with privacy concerns when it comes to consumers who may not want to be seen entering a business?
RC: We don’t have the ability to see live camera views. What we see is our algorithm monitoring cameras, asking the question each and every second: “Is there a gun in this image? Yes/ No.” If the software believes it sees a gun, it sends a screenshot to our operations center, which is staffed 24/7, 365 days a year by former military or former law enforcement professionals. They make the determination: “Is this, in fact, a lethal firearm? Yes/ No.” If it’s “Yes,” they press dispatch, which goes out to the building personnel and local law enforcement in as fast as three to five seconds.
The only thing we see is when the software believes it sees a gun; everything sits on the end user’s side of the firewall. As far as privacy goes, we are ultra-secure, whether it’s a Manhattan bank, a public school, or a cannabis business.
MJG: Do you have a story that you can share of ZeroEyes identifying a threat that a traditional surveillance system might have missed?
RC: We’ve had, I believe, 14 or 15 dispatches in the last 18 months that have led to arrests. One was at a subway station, when a man pulled a gun out and was pacing back and forth, waiting for somebody to get off of the subway. We dispatched the alert of him pulling out his pistol. He took off running as he heard the cops coming, and threw his gun in a trash can. With our information, the cops knew who they were looking for, what they were wearing, what kind of potential threat that person might be. They were able to track him down about a block away, and make an appropriate arrest based on the information that was dispatched to them.
MJG: How do you make sure your systems stay up to date with advances in the AI sector?
RC: Everything ZeroEyes does is in-house. All of our employees who determine whether or not they’re looking at a lethal firearm are former military or former law enforcement. We build our own algorithms, and train our own data sets in-house. We have a 5,000 foot green screen lab right next to our headquarters in Philadelphia. Our team walks around in front of hundreds of cameras every single day with guns of all different types. All of that data goes to our AI team to continue to train the ZeroEyes model, to progress and advance it forward.
MJG: What do you see happening with ZeroEyes in the future?
RC: I believe that at some point, visual gun detection technology will be like fire alarms. Hopefully we’re headed in the direction where it’s building code. Anywhere there’s a camera and a potential gun violence problem, ZeroEyes is a solution that can potentially help mitigate that threat.
