AI in Physical Security: Expert Success Stories Dispel Falsehoods

Hey, everyone. Welcome back to Innovation Obsessed, where we tell the stories of innovation heroes making the world a safer, better place. Today, we're bringing you something special, a crossover episode with the Secured podcast. In this episode, security experts, Mike Matronga and Mike Monsif bring insights from the trenches, sharing their real world experience from safeguarding schools and campuses to managing high traffic events. One example they share is how AI driven solutions like Ustaz Vision AI help detect abandoned individual at a graduation ceremony. This is from forty feet outside the perimeter, allowing security to intervene before any incident occurred. They also discussed the critical balance between using cutting edge technology and respecting privacy. They talk about policy. They talk about procedures. This episode is packed with insights and real success stories. So for anyone responsible for campus or facility security, it's a great episode. Let's dive right in. Welcome to the Secured podcast. This is, your host, Mike Matranga, with our cohost, Mike Monsit with ASAP Security Services. Today, we're gonna talk about the effectiveness yet the perceived dangers of artificial intelligence. And so with that said, I'll turn it over to Mike since he's a little more well versed in this than I am. I'm just gonna be here as more eye candy for you guys. Yeah. Certainly. Yeah. You don't wanna talk about that? This isn't your this is you know, you you I specifically wanna talk about it with you because you have been able to see how AI has changed security Mhmm. And and allowed you to be proactive in there. But we've really been dealing with AI and security for for a very, very long time, even before any of this other stuff from AI was there. And it was always always the big, you know, bad thing. Right? AI is is in there. And so if you go back to the very first versions of facial recognition, and today we don't even call it facial recognition because of all the negative connotations that are tied to it, the the very, very simplest first forms of of, facial recognition did have some biases in it. Right? And specifically to, darker complexion, brown and black, there were, there were some some, biases in that software. And today, we're still dealing with the damage that those first versions did, but it's not it's not done that way anymore. It hasn't been done that way by the major vendors who provide facial recognition. It has changed and evolved so much. And now it's it's truly a mathematical formula. Right? So that everybody's face is different and and so the the art the artificial intelligence, the camera, the the metadata, all of that underlying thing is actually performing a mathematical equation and based on the distances between your eyes, the bridge of your nose, your cheekbones, all of these different things that it takes into the measurements, it then calculates a number and assigns a numerical value to that person's face. And I think, you know, two thousand eighteen, two thousand nineteen, we rolled out, it was there was then a company called AnyVision, now called Oosto. Still a phenomenal product. If anybody's interested in in, that particular type of product, I would highly recommend that. No. I don't get paid by them. You know, two thousand eighteen, two thousand nineteen, us and Santa Fe ISD, Texas City ISD, Santa Fe ISD were the first to, bring that level of proactive facial recognition to schools. You guys were the ones who helped bring us, bring that into our campuses. And I can tell you, the early, variations of, of those, or that particular type of software or technology had did have such a negative connotation that I had to suffer through all of the, accusations of racially profiling. But, you know Here's what's crazy though. If I had to ask you, out of all the technology that we deployed That's definitely the best. And it was the most Effective. Proactive. Right? The most proactive. It was something that you saw somebody coming Yeah. Before they got to the front door. Absolutely. Everybody knew you were able to respond. I will tell you that is out of all of the technology that we had, and we had a lot back then, we kinda set the pace, which now a lot of the stuff is standard. Everybody thought I was crazy back in those days. And now the things that I did in two thousand eighteen are now common. And now I'm not so crazy. Right? I was just ahead of my time and because I understand the effectiveness of Proactivist. Yeah. And, I will tell you without a doubt, that was the most effective software I have ever used. And so if if given given the right software with, the right method of deployment Mhmm. And with the right people to, build that in a proactive manner, it it could be the only software you really need outside of just your general cameras and your access control. Right. You more than likely know who your threats are. You know, over ninety something percent. One time in a magazine or something interview. Yeah. But, you know, that kind of goes in one ear and out the other. Okay. Right. Multiple steps in the hen house. Wolf in the hen house. You kind of know who your people are that want to do harm to you or the people who that you're responsible for. And so if you you already have ninety five percent of the equation figured out, then you're utilizing this software proactively to identify those people who are going to come to you to Cause harm. Inflict harm on you. You're winning, man. You're absolutely winning. And so but, you know, I had to to wade through all the nonsense with all of the activists and the people who were looking at the early reports of AI, you know, facial recognition being, you know, discriminatory in nature against, African Americans and Hispanics when in reality, that's just not the case. The case is is that the software that we chose, we chose that because it didn't take into consider race or gender. Like you said a while ago, it is simply a mathematical equation. And so it I used to laugh when people would get all bent out of shape. I'm like, what are you what are you upset about? It's taking a measurement of your face. It was. You know, we've met some pretty amazing people, like, on our journeys through schools and everything else. And one area that continues to amaze me is the front office. Office. Oh, yeah. Oh, man. The brain of the school, man. The truancy side, that right there, they, by sight, know parents, you know, families that are divorced and have remarried both sides. Yeah. Or both going through divorce. And so, you know, when we've looked at schools that didn't have a visitor management or didn't really have a process to process students in or out, You know, I've sat down with, some of these individuals and been like, okay. But if you needed to know, like, who picked up Johnny early Yeah. At, you know, at two o'clock on this day, it'd be like, I could tell you. Anyway, hold on. There's no you have how many students here? But, yeah, but I could tell you who who came in. But it's not written down. There's not a process. No. There's no I know. And and it it's true. This tool gives them so much more in there, and the reality is some days one of those people is gonna be out. Right? And you gotta be able to give that that school or that business, you know, let's not just leave businesses out of here a fighting chance. You know, we've successfully rolled this out within, corporations, right, and corporate wide. And when they actually engage and start running it, Right? We get through all the legal hoops and then they roll this out, and then that first thing happens, hey, we have a person that's doing espionage and we think this is a a an associated actor or, we have an individual who's made a a a a former spouse that has made a threat to harm or injure this individual Who's an employee. Who's an employee here. We were concerned terminated employee who's made threats. And when we start working with these corporations and or schools and we get the processes built and they start using it, how it changes their security culture. And out of everything that I think that we sell and recommend, this is one of the things that we have to apply the most pressure to initially get the buy in. Mhmm. But then after it's deployed, it's one of those things that they're the lifelong customer. There's somebody who's there like Yeah. Oh my gosh. You have done this. You've you've gotten this done and done this and everything else. But we had to try so hard to get it installed to begin with. Right? Even though we got it installed and it was proven to work, the school district that I was formerly an executive director of school, of safety security at, school safety and security at, they no longer use it. You know why? Because the superintendent gave in to the social pressures of of in, you know, inaccurate information Right. That was presented by a group of activists who had a personal grievance with me, and they gave into it. And they got away from it. And it was the most effective tool that we've ever used. Well and and we've seen situations where, let's say, an organization has a law enforcement component or they have services on there, and they just absolutely will not discuss it No. One hundred percent. Like, we will not talk about facial recognition so much so that we don't even call it facial recognition anymore. Oh, yeah. We've gotten to where we have to call it image likeness because people get so weirded out. But, you know, it's what I don't understand is this. We're not entering you into the system. We're entering known actors, people that we know want to cause us harm. An employee, like you said, who has a spouse who's caused a threat, a former employee who's been terminated due to whatever behaviors that's exhibited signs or symptoms of pre attack, who's, you know, exhibited that they have an unusual interest towards your facility or an another employee, you know, a group of individuals who may be, anti whatever it is that you're doing. A ring of theft. Right? People coming in there. The the organized crime coming in. You know, when I started my career in security and actually in in the installing security system side, I did a lot in retail. Right? So we every major retailer that in the nation, we did we did work for. And you would go into a loss prevention room Mhmm. Or in a manager's office, and it was often pictures taped up all over the walls of be on the lookout. Right? BOLOs. Yeah. Be on the lookout. Right? Putting out this big book that's this thick. Yes. You know, it's like And how do you seen that? And so How do you recognize all these people in a book that's this thick? Yeah. And these these police officers, security guards are supposed to study this book and from memory recognize this. And if you think about a retail environment, right, that that this is in there, this loss prevention person is supposed to know Yeah. Just from sight that this person came in because they've inflicted so much pain upon the organization in theft and damages, and it's truly on the memory recollection of that individual to know. Yeah. Whereas, I am not saying that technology and and image likeness, facial recognition is imperfect. But what it does is it gives you a jumping off point. Well Okay. And so, hey, you just got a ping that this person's on here and we think that it matches. Yeah. Okay. Well, now I'm gonna pay attention to that individual. It may or may not be. And me as a human, I'm gonna make a decision. Well, it's incumbent upon you to write the policies that that don't allow for any type of, you know, I wouldn't say don't allow false positives because you're gonna you're going to have false positives. I don't care what system it is. Sure. You're going to have false positives. It's incumbent upon you as the person responsible for that software to write a policy that states that we're not gonna solely rely upon the software to identify. We are going to use a series or a set of human eyes or multiple sets of human eyes, and that policy has to be written and adhered to so that you don't infringe upon someone's rights. I treat it just like any other source of intel. Right? You you you get fed information. Yeah. And you need to look at it. You need to decipher it. You need to validate it. And is is this real? Is this actionable? Is this is this, you know, something that I need to do something with? Right. And and it doesn't matter whether that's a tip coming in. Right? Or it's a ping that you may have somebody as a match in your in your database. I mean, think about it. Right? Let's take, think let's think of, let's think, what could it could have been any of them. Parkland, Uvalde, Covenant Christian, Sandy Hook. Someone knew that those individuals had some type of unusual interest or infatuation. Right? If school districts now I'm not saying you just go put random people into your system, but if if there's a if there's a concern, a valid concern that this person may wanna do harm, maybe they do, maybe they don't, but they've exhibited that. Right? We put them into the system. That person then makes their way upon to the campus. The system identifies. You have a series of notifications that come out. Not saying that it would have prevented anything in Sandy Hook or Parkland or any of that other stuff, but what I will tell you is from my own personal experience, I have seen it work that way Right. Because we had a robust system that identified students that either had law enforcement issues that they were not allowed on campus, or they had exhibited, threatening behavior towards our campus. Real real world search situation. Right? So this year, specifically k. We've had the proof of all of the shootings at sporting events. Outdoor sporting events, specifically football games was a lot of them. But right? And so some of that, if you go back and do the research, was a spillover. Something happened in the school, right, during school hours. Maybe there was a fight. Maybe there was an altercation. Maybe somebody was suspended or expelled. Friday night rolls around, and oh, we're gonna have this football game. Well, all of a sudden now, it's free game. Oh, it's a student. We're gonna let them in. Right? Yeah. Not that that student was just expelled for fighting over somebody else or something happened, and now he's gonna show up in front of everybody Mhmm. And and and act the big man and do something about it. Right? If you had deployed, We did deploy. Facial recognition. Yeah. Just talk about hypothetically. Let's go back. We'll go back to ours. If you had deployed it and this person was expelled or suspended and the policy is that you're not allowed to attend extracurricular activities. Right? Mhmm. Because you were suspended or expelled, they could be added to a list. And when they're going through a security checkpoint to get into the football game, right then and there, it lets the people know the social supposed to be there. Can I see your ID? You match the description. Okay. You know you're not supposed to be here. Right? You you you stop that right then and there in its tracks. May of two thousand nineteen, we deployed facial recognition. You guys designed the exterior boxes that we installed on every entry gate. And and in in the final hours, it was In the final hours. But we got it. Cellular. Now now now I had just hired an individual to be my director. Right. I wanted to see it work. And and so we deployed it. Mike and his team designed the boxes to house the cameras that we installed on a wrought iron fence and to check them out. Up temporarily. Put them up temporarily. I had two of my team members working in the press box as Overwatch monitoring the system because we knew we would have people enter our facility or enter our graduation that was not supposed to be there. Yep. And I remember having that distinct conversation with the former director and you just saying, I just wanna see it work. I I just I if it works tonight and someone get you know, someone attempts to get in that's not supposed to be here and it catches them, then I'll be bought in. And sure enough, we had a student that was a violent student that had been sent to the alternatives campus, was not allowed to participate in any of these particular types of things. The system identified him forty feet away from the entry gate. We got a notification. By the time we walked to that side of the field, he had already walked we let him walk in. In that time, we were processing the policy validating the policy. Two sets of eyes, three sets of eyes need to identify along with the software, is this the person? So my guys up top are going through. Yep. I think that's him. Yes. I think that's him. Officer, can you check this out? Yes. I think that's him. We let him walk in. He sits down. The second he sits down, the officer taps him on his shoulder. I need to see your identification. You're not supposed to be here. The first thing out of his mouth, how did you know I was here? Yep. He thought he was just gonna slide right in under the radar. He did. And we got him. You know, when you do what we do. And then many times, we deploy technology and we hear about the success. Very rarely, you know, because we deploy it and then something happens and then success, and we hear about it from the customer and they're excited. Mhmm. But very rarely do we get to, like, witness the customer's reaction in real time. And and and that was that was really a a Well, we always I mean, but I mean, you I was client. You were or I was customer. You were vendor at that time. Yes. And, you know, here we are five years later. Now we're, you know, working together, business partners, essentially. And, I mean, it's just the the level of professionalism, the can do attitude, the no fail mission. Those are the things that I'm used to. You guys you guys rose above and beyond and rose to the occasion for that and made sure that we that we got what we paid for. And, you know, I tell you, man, I was a believer after that. What was also fun about that is we had you had a a significant amount of law enforcement agencies there frozen that that night. And, so we had an another individual that had tried to come in one gate and stopped stopped him outside the gate, and he turned around and left. What did he do? Went and put his gun in his car. Well, now he went to the other gate. Tried to get in the other gate too. Took a shirt off, put a different shirt on, came back Got him again. Got him again. Got him again. Different gate. I remember the chief the police department down there going, I can't believe it. Right? It's like, this is he's like, this is amazing. It works, man. I tell you. And so then after that, we just had win after win after win after win with that software where, you know, there was incidents where someone wasn't supposed to be there or there was a restraining order, and officers were notified. You had the process in place. Your team notified the officers on-site. They went out into the parking lot and identified the problem out there Yeah. And handled it before they even approached the front door. You know? You know what? I think that's the that's the thing that we need to differentiate is that there's artificial intelligence. It's just such a broad term. Right? It is. And unfortunately, everything's getting grouped in. Everything's getting grouped in as it's bad. Right? Listen. Anything you if you eat enough cheese, it's gonna be bad for you. If you eat too much drink too much soda, it's gonna be bad for you. Right? If you if you abuse artificial intelligence, it's gonna be bad for you. Right? There's it's Well, it it If not done right, it can be a problem. It can be abused. Some of the things that we're seeing today that are changing, and and and this is why I really wanted to talk about this, was that some of this as soon as you use the word AI with security, people automatically start thinking of, like, chat g p t Yeah. Or the meme videos where you can make anybody, somebody who's been deceased for twenty years, you can take a photo of them and make them say something in a video that looks real.