AssociationHelpNow

Facial Recognition in HOAs and Condos: Access Control, Privacy, Insurance & Legal Risks

• Raymond Dickey

Share episode

Use Left/Right to seek, Home/End to jump to start or end. Hold shift to jump forward or backward.

0:00 | 1:00:20

Facial recognition is no longer a future issue for condos, HOAs, high-rises, and community associations — it is already here.

YouTube: https://youtu.be/g-KmC6A2gTc

CEU video will be available on Jun 30, 2026 12:00 AM EDT at: https://attendee.gotowebinar.com/register/1688654602841957467 

In this discussion, our panel looks at how facial recognition and AI-powered camera systems are being used for access control, license plate recognition, amenity access, security alerts, and resident convenience. But the conversation also goes deeper into the risks: biometric data, privacy concerns, discrimination claims, vendor liability, cyber insurance, police access, data retention, and whether boards should adopt this technology at all.

Topics include:

• How facial recognition actually works
• Why it is more than simple face measurements
• Access control for buildings, pools, gyms, gates, and restricted areas
• Whether masks, sunglasses, makeup, or photos can fool the technology
• Why boards should question “99% accurate” sales pitches
• Biometric data, cyber insurance, exclusions, and vendor responsibility
• Discrimination and privacy concerns
• Police requests, subpoenas, and community policies
• Why boards need written policies before adopting the technology
• What questions managers and boards should ask vendors
• Why facial recognition may only be the beginning of AI-powered community surveillance

Panelists:

Dawn Becker-Durnin, CIRMS
Acrisure
dbecker-durnin@acrisure.com
www.acrisure.com

David Byrne, Esq.
Ansell Grimm & Aaron, PC
dbyrne@ansell.law
www.ansell.law

Raymond Dickey
AssociationHelpNow.com
www.AssociationHelpNow.com

Gregg V. Gerelli
Gerelli Insurance Agency, Inc.
gregg@gerelli-insurance.com
www.gerelli-insurance.com

Sean A. O’Connor, Esq., CCAL
Clarkson McAlonis & O’Connor, P.C.
soconnor@cmolawpc.com
www.cmolawpc.com

John Tani Jr.
Elevance Health

This content does not constitute professional advice.

#HOA #CondoAssociation #FacialRecognition #AccessControl #CyberInsurance #CommunityAssociations #HOAManagement #CondoManagement #AssociationHelpNow #BiometricData #Privacy #CommunityAssociationLaw

SPEAKER_05

Hi, Ray Dickey from Association Up Now, South Carolina and Hudson Valley CI. Thank you so much for being here today. This is a pre-recorded session, so I don't think we have anyone from the audience here. That being said, panel, don't say anything that you don't want anyone else to hear. Don't count on me to delete things, Sean. You never know. I could make a mistake. Today's not the day to tell us any big secrets or anything. All right. Today we are very lucky to have John here, who's going to introduce himself here in a second. We're going to talk about facial recognition. Um, John, this isn't the future, it's here now, right?

SPEAKER_03

100%.

unknown

Okay.

SPEAKER_03

100%. And it's getting better every day.

SPEAKER_05

So we're going to get all your questions answered. You can use your question feature. I actually still get those questions even during the pre-recorded session. You just may not get a response for a couple days until I run the report. If you're here for CEUs, you're watching the pre-recorded session, you need to be using the go to webinar system. If you're not here for CEUs, you don't need to worry about it. You will be prompted to enter using the question feature CEU. And you have to do that about three times. And that way I can let the educational organizations we work with know that you are here. So keep an eye on that. Find your question feature and make sure you're on our go-to webinar system. Otherwise, you may be watching on YouTube or anything like that. And if you don't need CEUs, that is absolutely cool. I love comments and I love questions. Even though I'm not live, send it. I do check it out and I do get back to people. Thank you so much. With that being said, I'm going to have our panel introduce themselves with who I see first. And that is our great guest, John.

SPEAKER_03

Hi, so my name is John Tanney, and uh I recently retired uh after 26 years with the United States Secret Service. And um my career spanned everywhere between New York City, Washington, D.C., Honolulu, Hawaii, and you name it. Uh by the time I had retired, I spent uh the better part of that career in our technical security division and looking at a lot of uh technical solutions to problems that we were facing. Um so now I work for a very large company uh in the corporate world uh doing domestic security operations. So super happy to be here uh just to point out whatever I talk about today. I'm not gonna give you any secrets. I'm not gonna talk about what I really used to do. Um, and none of my views actually um are supported by anybody that I used to or currently work for. So this is just me talking to you guys.

SPEAKER_05

Oh, we're gonna ask you a lot of top secret questions. No, we're not gonna do any of that. I do have let me get a couple secret service questions out of the way, John. The same ones you know, Tom. Tom, he's been on here many, many times. Have you ever worn your sunglasses at night? Because Tom, well, because you're in the Secret Service, we know it's a thing. Do you wear sunglasses at night, Sto?

SPEAKER_03

Yeah, no, don't do that. It's uh it gets harder and harder to see for sure.

SPEAKER_05

You ever saved a pop star and carried her out of a concert and there's music playing in the background? Have you ever saved the pop star in your career?

SPEAKER_03

No, but good reference. Uh, but there are lots of good stories with lots of good pop stars for sure.

SPEAKER_04

Have you ever guarded have you ever guarded Whitney Houston?

SPEAKER_03

I have not. I made another good reference.

SPEAKER_04

I'm just wondering if you're not gonna be able to that's what I would Dave.

SPEAKER_05

That's what I was talking about.

SPEAKER_04

That was it in my wasn't that different movie? Was that called My Bodyguard?

SPEAKER_05

John knew which one I was talking about. I knew I knew where you were going. And last question, John. And have you ever saved the White House from impending doom? Like you were the only man that could do it. Everyone else was unavailable, and only you could do it. And you save the White House from impending doom. Have you ever done that?

SPEAKER_03

Uh, have not done that. Um, but I came close to ordering baby peas in a grocery store once, if you understand that reference.

SPEAKER_05

No, I don't.

SPEAKER_03

I don't I have tests, great movie. Guarding test, great movie.

SPEAKER_05

See, you do know all the you do all you guys do know all the Secret Service movies.

SPEAKER_03

Okay, cool. I know them all, and I will tell everybody this. If you still want the most accurate Secret Service movie today, in the line of fire, Glenn Eastwood 100%. Oh, that that's cool.

SPEAKER_05

I'm glad it's him too. Uh by the way, Tom did answer that he did wear his sunglasses at dusk, but he denied ever saving a pop star, and he said he did not single-handedly ever save the White House. So you're on par with the time's results. Okay, uh, let's go to who I see next, and that is Sean.

SPEAKER_01

Good morning. My name is Sean O'Connor. I'm an attorney. I'm a shareholder with the law firm of Clarkson McIntosh and O'Connor. Our office is in Mount Pleasant, South Carolina. We have a statewide community association law practice serving all areas of South Carolina. I'm a longtime member of the South Carolina chapter, been a member of the chapter since 2008. Uh, have served as uh president of the chapter in 2023 and uh many years on the board. Um, serve on the uh blue ribbon panel for the South Carolina CAI, and I'm also a member of the CAI National Business Partners Council, and I'm a C Cal.

SPEAKER_05

Anything else?

SPEAKER_01

That'll cover it for now. Okay, thank you.

SPEAKER_02

Uh Greg. That's hard to follow, right? I don't know if I can keep up with Sean here. Um, Greg Girelli with Jirelli Insurance Agency. We're located in Cold Spring, New York, and we do insurance master policy programs for community associations throughout the Hudson Valley and surrounding tri-state area, and we're on the board of the local Hudson Valley CAI chapter. All right. Uh, Dave.

SPEAKER_04

Good morning. Uh Dave Byrne from Ansel uh Law. I'm the um uh chairperson of my law firm's community association practice, and we represent uh HOA's condominiums and cooperatives in New Jersey, New York, and Pennsylvania.

SPEAKER_05

And let's make sure we leave time, Dave, for you to give us your personal opinion on how you feel about surveillance. Because I know you're gonna I know you're gonna have a good one. We'll save that for the we'll save that for last.

SPEAKER_04

I hate that, but I do like the fact that I can get into my um uh online banking app just by looking at the phone. I have to say I do like that.

SPEAKER_05

Okay, well, there you go. I'm surprised you said that, but good good for you, Dawn.

SPEAKER_00

Hi, everyone. My name is Don Becker Dernan, and I'm a firms insurance agent, and I work on behalf of Acrashore. We are a national as well as international retail insurance uh brokerage, and we provide employee benefits as well as commercial property and casualty, which is the division I specialize in for condos and HOAs.

SPEAKER_05

Okay, so John, it's here. People don't think I don't think people realize that facial recognition is already here and it's already being used, and they've already been part of it being used.

SPEAKER_03

Yeah, and um, you know, to be quite honest, it's actually an older technology. Um, it has been around probably for you know 15, maybe even 20 years. Um one of the one of the the best uh facial recognition algorithms that existed uh started very on, early on. Um it's called PitPack Tech PitPat Technology. Um in 2011 it was sold to Google, but that originated out of the Carnegie Mellon. Um, and it stands for um Pittsburgh Pattern Recognition. And so that's been around for a very, very long time. And um machine learning, AI, it's all making it a lot better than it used to be.

SPEAKER_05

So there it's already in play at some associations I know are already using it. It's they're mostly expensive, high-rises and things like that. Um, what are some well, first, how does it work? I mean, it's kind of an obvious question. We don't want to dive too deep, but basically, how does it work for those that don't have any idea?

SPEAKER_03

Sure. Yeah, facial recognition, it's it's a way of using software to determine the similarity between two faces in order to evaluate a claim, right? And so, like you said, it can be used for a variety of different purposes. You can either use it to sign into your phone or um uh search a database, right? So um, and there's a lot of um, I think, misconception when people think that they know how it works. Um, people think that it actually takes measurements of your face and then it stores it in a database. That's not actually what it's doing. It used to do that. Um now it's you uh with AI, machine learning, deep neural networks, it's actually using your face to create um a pixel map of your face, right? Uh computers think of binary numbers, right? They think of zero and ones. They don't see pictures like we do. So they have to convert your picture, your face, into a series of zeros and ones, and they do that through shading and through pixels. And that that's important because when you look at how accurate these systems are, it really depends on the algorithm that you're using and also the camera, camera placement, and that sort of thing.

SPEAKER_05

So I thought I knew how it worked, and I did not. So, how does it work then? Does it work with the coloring of your face? Let's say I go in and get a heavy tan and I don't. What I'm so interested. How do how what does it what do the pixels include?

SPEAKER_03

Yeah, so this can get like super complicated, and I don't want to dive too far into the technology. There's some great resources that I can I can share with people after the fact. But basically, it um if you've ever gone into a photo and used like Photoshop, right? And you you zoom in as far as you can zoom, you see these little squares, right? And so pixels are the they stand for like right on PIC, and and it is the smallest unit that you can get in uh in a digital photograph. And so it will take those those picks and it will assign a number to them. So think of the darkest black as a zero and the whitest white is like 500, right? And in a picture, the shades of your face, your eyes, your eyebrows, all of that, and they all associate a pick to a number. And then it starts to build basically um a template of what your face should look like. And then it compares that, it runs a bunch of filters through it, and it compares it to if you have a database, um, you know, that template in other databases. So it can get pretty confusing. Um, but like again, there's some really good resources, it's really interesting. Just when people say, Oh, it's just taking measurements between your eyes and your nose and your mouth, that's it's it's way more complicated than that.

SPEAKER_05

So I didn't want to like go in this, and if you don't want to answer a question, by all means don't, because you have a responsibility as part of your job not to tell people sometimes that what works. But what happens if I put makeup all over my face like a clown? I put stripes on my face or something like that. Do you mind answering that question? Would that impact it or or would not?

SPEAKER_03

Uh today, no, it really doesn't. As a matter of fact, um, the National Institute for Standards and Technology, which is right down the street from me, uh, my kids actually used to go there for um for uh for for preschool, right? So they they do these tests and they've done uh like years-long tests to validate the algorithms that are basically making facial recognition work. And so it's kind of like a test, right? So they did one in 2017, they they're currently doing one in 2024. And so what they found is that things are getting so so good that recent tests, like as of 2024, if you had 70% of your face obscured, like some of the best algorithms are still getting 97% true positive rates, which is pretty amazing, right? So that means you can wear a mask, you can wear a hat, you can wear sunglasses. And sunglasses are really interesting. I mean, going back to the theme, right? Um, that you're actually able to open up your iPhone and other things with sunglasses, uh, not because of the algorithm, but because of the technology that's actually built into the phone and how that translates through uh your sunglasses. So it like I said, it can get really complicated. I just want people to realize it's not as simple as just you know taking vector measurements between eyes, nose, and mouth, because with masks, obviously that would be pretty difficult.

SPEAKER_05

Wow. Um like I said, I didn't know any of that. So that's pretty exciting. Um, it can work from quite a distance too, right? Like how far away can you be? Does it depend on the camera? Can you give us some idea of that?

SPEAKER_03

Yeah, so it it can, but what you're trying to do, right, is when you're looking at accuracy, you're really trying to figure out two things, right? You can have a facial recognition technology that basically is you you want it to allow you in, right, when when you're authorized, and you also want to keep people out when they're not authorized. Okay, and your phone kind of does the same thing. So there's a trade-off between the errors that you get when you you like you open up your phone and sometimes it doesn't recognize you all the time. Maybe it takes two or three. And they built that in because they don't want Joe next door picking up your phone and getting in. So there's a way to kind of tweak those values. But when you talk about camera placement and distances, it all uh it all blends into the piece of the puzzle. The most important is the algorithm that's being used. And then, second, um, experts have said camera placement is most important. You want high fidelity cameras, right? A lot of megapixels. We talked about pixels and why that's important, but then it also needs to be placed in a certain way where it is actually picking up a really good photo view. Not to be said, I have seen technology that will get a glancing photo of people, it'll get their side view, it'll get just a little piece of their face, and it's actually matching. But you want to set yourself up for success. So if you have the ability to install cameras and place them in strategic locations, that's what you're gonna want to do.

SPEAKER_05

And people like it because it's a convenience at these high-rises. Like I said, they're using them all the time. You don't have to get anything out of your pocket. You can basically, it just sees you and it lets you in certain doors or not. Um, I know there's even technology now. I think they're selling it. Maybe, Dawn, you've heard of it, where um I think they they can actually see people in cars driving into associations. Um, Dawn's nodding, everybody, yeah, everybody's agreeing. John, that's amazing too, right? Associations are some are using it right now where people are driving in with cars and it's picking up who they are. Is that true or false?

SPEAKER_03

Yeah, that's not hard to do. It's not hard to do at all. What I would think, what I would throw back on it, throw back on you all, because I I'm really uncertain as to how it's actually being used in in the environment, although you've given me some clues, right? Looks looks like or sounds like you're doing it for access control. So uh I would I would be cautious with that. Um, I would also not just want to rely on that. Uh so a lot of times people, and uh I know somebody doesn't like the privacy here, uh privacy implications of this, but they're using that in in combination with uh license plate readers, right? So you're getting two forms of authentication, um, or maybe even a key fob or whatnot. Um and so yes, that that is it, it's an access control piece. So it really all it's doing is saying yes or no, right? And and I think um, you know, for the viewers, you have to figure out what is your tolerance for either errors uh on both sides.

SPEAKER_05

So in our sorry about that, in our industry, they would be using it for access control into the let's say it's a building, into the building. Maybe you are allowed to have access to the pool, maybe you're allowed to have access to the gym, maybe you uh maybe you're on the board of directors and they want to allow you to have access to certain maintenance rooms and things like that. So they would be using that technology fully for access control and tailoring it based on which each person should be allowed, what area they should be allowed to visit and not visit? Um, does anybody in the panel think I'm forgetting any kind of aspect in our industry that they may be using facial recognition for? Okay. All right. So, you know, before we dive into a couple other deeper parts of this, John, I want to give the panel a chance to ask some questions because a lot of them were really interested. They may not have any. Sean, any technical questions?

SPEAKER_01

All right, no. Okay.

SPEAKER_05

Any anyone have any additional technical questions? Or did I cover? Go ahead, Don.

SPEAKER_02

Not one. Oh, I was just curious, on can you fake the facial recognition with either a mask or a computer screen in front of it where it was looking at the same pixels and be able to get you access to something?

SPEAKER_03

So it's a good question. A mask, obviously. Yeah. Yes. Well, it depends on what type of mask. Are you talking full face mask? Yeah, obviously, right? Because that's not your face. Um, if you're just wearing like a face covering, um, it it it had like the algorithm. If you have a good algorithm, it will it will not even register that. It'll it'll measure off of other elements of your face to get you to what it thinks is a is a is a close match. I have seen some things online and I haven't verified them as um like when I was a kid and we would uh dress up as robbers for um for uh for Halloween, we put the pantyhose over our face. Um I've actually seen people printing faces on the pantyhose so that when you put it over your face, you're getting facial structure, you're just changing it. I I don't know. Um I think anything that has a technical component can be faked, although the bar is getting smaller as to what you can get away with.

SPEAKER_04

Are are like cameras that are installed generally like ATM cameras or cameras at like you know, Penn Station in New York, do they have this kind of technology embedded in them?

SPEAKER_03

So I I don't know what Penn Station is using. Um I don't know what every ATM is using. I will say this. Um there's two types of really when you're looking at facial recognition, right? You're really looking at two use cases. You're looking at verification, which is one-to-one, which I assume maybe is what you guys are doing, right? You have a database or you're trying to match one person to another person. The other one is one to many, which is identification, right? And identification is very hard, harder than one-to-one matching. Think of Penn Station, thousands of people rolling through there. You're looking for, sorry if anyone's name Joe, I just use it generic, Joe or Jane who just committed a crime, right? That is a lot harder because the computational process and algorithms need to be taking that face that you have and comparing it to everything that it sees all the time, super fast, right? Um, that's harder to do than kind of with what you know access control is. Access control is is generally very easy because you have a whitelist, you have people, uh probably a small whitelist. Um, so that's easier. The interesting thing with ATMs is that there are some studies. I missed did the study and they found that ATMs are not very good at facial recognition, or some of their algorithms are really bad because the quality of photo and actually the placement of the photo, because it's not up high, it's it's low in a lot of cases. And uh the quality and consistency of the photo when it's taking it is what's really also important.

SPEAKER_05

Uh hold one second, Don. Um, Dave, by the way, they're already using it. I think it's in Madison Square Garden. You didn't read any of those stories, I think, where the owner has placed it. Yeah, Greg's nodding his head. Hate to tell you something, Dave. It's already being used out there, and there's been some lawsuits around it. But yeah, I think one of the major right, what where was it, Greg? You were nodding your head. You know which one I'm talking about.

SPEAKER_02

I think I heard about that out of Madison Square Garden. I think you got it right. Yeah.

SPEAKER_04

And he's saying, but John's saying it's very difficult to match to you know, find um, you know, I don't know, Khalich Sheikh Muhammad walking through the turnstiles in a Madison Square Garden, uh, just based on a description that the police may have from a witness that called in an hour ago, right? That kind of thing. I mean, you wouldn't be able to do it based on a description, but John, you could do it based on like how quickly is this stuff loaded up. You know, you it's gotta be really difficult to identify a random person out of thousands and thousands of people.

SPEAKER_03

Well, so I don't want to misspeak. I don't want to say it's so it's so identification is harder than verification, that if there's a distinction there, right? And so um it depends on the quality of the photo that you can get, and you can absolutely do it, right? So I go back to the pick any born movie, right, that you that you want to reference, and they're running around Europe, right? And they're trying to get through the different countries, and they have their face and they have it loaded, these digital green faces that came from like a fax machine, right? And they have it loaded into their database, and they're like, oh, they were over here and they're over here. And I used to laugh because I was like, that's no way, like that's not even possible. Now, fast forward, you know, 15, 20 years, and um, if you look at the Boston bomber uh case, right? That's a little bit different. They were, I believe, I don't know 100% sure, they were using a little bit different technology, but they were still doing the same thing, they were still pattern matching off of faces, right? And so it is when you're doing identification, the fail-safe is that the computer and the algorithm can give you a clue, a pretty good clue with a confidence rating, but then really a human in the loot needs to go in and say, all right, is this truly this person? And you see that a lot in uh local police departments. Um, I've been watching a video or I've been watching a Netflix series called uh Homicide Squad based in Louisiana, and they are not making any positive identifications off of uh any facial recognition systems they have there. If they're getting clues, they're actually making positive facial rec facial identification off of either people who know the subject, um, investigators. Sort of thing, so I think there's really this shift to say, hey, like if we're gonna use it for identification, we also need a human in the loop to make sure that that's who they say they are.

SPEAKER_04

It's like a total crutch. Like you can you watch like a crime show and they're just like, okay, that's just a lot, let's just hack uh the subway cameras and we'll find the guy in two seconds.

SPEAKER_03

All right.

SPEAKER_05

Yeah, I don't think that's happening. So, Dawn, you had a question.

SPEAKER_00

I don't know if it was a question, but on the Madison Square Garden, I think it's important, and it's gonna parlay into what we're talking about. That lawsuit was about banning individuals from Madison Square Garden. And the technology was used to enforce people who they didn't want to be in the stadium, and because it was private property, the judge did wind up dismissing that lawsuit. It was a class action lawsuit. But I think that's in our condo and HOA space, really important factors. Are we now going to use that information to not allow people? And what if there's a mistake? So, from an insurance perspective, we talk about this quite a bit because we're funding a lot of that litigation. But um, I think your audience will probably want to know in the context of legalities, what can they do? Or if it misidentifies someone incorrectly and now they can't get in, what do we do then?

SPEAKER_05

We're gonna get there, Don. Don't worry. Um, I think what it is too, John, for our industry, for the most part, they'd be using it. You know, you can't get into the pool, you can't get into the sorry, it lets you in something compared to not letting you in, right? Most places are gonna have a person available to correct that. Just like if my key fob doesn't work here at the building, it doesn't mean I can't get in, right? There's gonna be a person also at the front desk. But if I have my key fob, it it'll open the door for me more quickly. Um, you had spoken about confidence scores really quick. You know, for people looking at to purchase this technology, right, John? And we're gonna talk about the legalities, but we could just jump right to this part. They just want it for access control. It's available now, right? Because it could be purchased now, correct?

SPEAKER_03

Yes, yeah. And so here's the thing that I would say if you are um if you're looking at installing this, right? If you look at making the investment, this is no different than what I used to do with the government and looking at technology and for it for investments. First, you need to establish your requirement. What is it that you are trying to do, right? And then you need to go out and a lot of times people will come to you, but then you kind of go shopping, right? And you take that requirement statement, and then you compare with what you really need to what that um what that software is offering. And you're gonna get a lot of snake, snake oil salesmen. They're gonna claim 99% accuracy rates. When people start saying 99% accuracy rates, like you kind of that's where you dig in, you dive in, and you start doing your research. You want to know what algorithm are they using? Because a lot of these uh third-party vendors they didn't develop their algorithm. Some of these algorithms have been in development for 30 years and they just sell them to you know a mama-pop software. So you need to know which one they're using, and once you find that out, then you can really understand how accurate they are. Um, like I said, NIST did a study. I think they have the top 30 algorithms that uh they have proven as and they've ranked them as like is the best one. Um, and so that's what you should be looking for. Um some of the software out there, they will vendors will come in and have had this done. They'll say, Well, we don't even return a match unless it's you know, we're 70% confident that we have a match. Um, other vendors will allow you to kind of slide and set those confidence scores to the point where you're more concerned about uh keeping people out that shouldn't be, kind of like your banking app. Or are you more are you more concerned about you know accidentally not letting somebody in that should be? So um there's really a lot that kind of goes into it when you're selecting them.

SPEAKER_05

I think associations could they could save a bundle with this technology. Obviously, we have to go with the legal and insurance aspects because they're not gonna need key fobs. They're not, they're not gonna, you know, um, people aren't gonna be able to use someone else's key fob to get in and things like that. The technology is gonna be there and it's only gonna let people in that are supposed to be in compared to someone else sneaking in. It'll even keep people from piggybacking in, theoretically, too, right, John, because that's an issue too. People, one person used the key fob, they slide in, and then four people slide in after them. This will just like at the TSA, this'll technology will catch those three other people piggybacking and probably set an alarm off or something. I mean, that's all that's all here right now, right?

SPEAKER_03

It can, and that's that goes more right. That that expands like just facial recognition to really artificial intelligent algorithms that are detecting anything. Um, I have seen instances where offices have these algorithms running on their security cameras and they have picked up fires that are occurring. Uh, that somebody somebody walks by a door, an exterior door, and starts a fire, right? And it flags it as arson. I've seen it where um, you know, you might have uh a water break from a really cold day, you know, New York, Michigan, wherever, and your sprinkler head pops, and now all of a sudden all that water is flowing into your office space. Um, I've seen where it can alert all that so that you can get people there to clean that up. Um, so it's really moving away from facial, I mean, not moving away, but like adding all of these other elements to it. You can stop piggybacking. Um, well, not stop it, you can get alerted to it. But um, you know, you the problem with uh key fobs versus facial recognition is all in the backbone, how they're transmitting that data to your server. So if all of your cameras are running on Wi-Fi, probably not the best uh because Wi-Fi can be interrupted as we were talking about earlier. Um, it can be corrupted, yeah, signals can go bad. But if you're using hardwire power over Ethernet for your cameras, that's pretty good. The cost of cameras, um, it was coming down, but with this AI splurge and all of the chips that are like cameras that that you need, the high megapixel cameras, that cost is increasing. So it's really kind of trade-off where you know your key fob uh access control is mostly all hardwired in. It's uh technology that's super mature. And so maybe you know, if you're looking at a solution, you want some redundancy there.

SPEAKER_05

All right. So at the very end, I want to jump back, John. If you could think about it while we're talking about the insurance and legal stuff, stay by all, but think about questions that boards should ask. Because you I didn't get a chance to talk to you before you came on here. So I think you have an idea of what our industry is looking for in this regard, right? Did we give you all give you enough info on that? So I'm gonna jump back.

SPEAKER_03

Absolutely.

SPEAKER_05

Yeah, I'm gonna jump back to you. And what are some questions that a board or a manager or a management company should ask if they're interested in this technology? Who should they call? What should they ask vendors? Should they get referrals? What kind of technology, whatever you feel comfortable sharing. All right, so I'm gonna let Dawn go because as everyone knows, Dawn loves insurance. She gets like super excited. And you know, Greg, Dawn loves this whole topic of data, right? She's totally into it. And Dawn's the fence, and Dawn's the fence. It's her job, and she sees all the lawsuits and the trouble that come around it. All right, Dawn, I'm gonna let you get the ball rolling. And what are okay?

SPEAKER_00

So, my two things when I'm speaking to a board and a management company about information and data is how are you getting it and where are you keeping it or storing it to protect it? The maintenance of documents, I find that there's a huge jump off for a lot of our boards, meaning that I am not seeing a quality control in an administrative function to keep documents and records secure, nor do boards seem to have that transition to understand where those documents were, especially if it's a new management company, new board member. So a lot of our insurance policies, unfortunately, will have exclusions, and Greg and I are going to get into that in a little bit. Um, because the carriers are seeing these lawsuits, they're seeing the payouts, and they're large. So I think the focus that I'd like to see some of our boards and managers do is to prevent the losses from occurring before the insurance gets into um the equation.

SPEAKER_05

Greg, this is a, you know, I think people are finally getting turned on to, you know, um cyber insurance and and things like that. I think we're finally turning the corner, are we?

SPEAKER_02

Yes. I mean, more and more boards are paying attention to it when we start talking about cyber insurance and the proposals, we're understanding it better. There's been more court cases, so there's more examples to go over to explain how things have happened to associations. Facial recognition at the moment isn't one of them. But like Dawn was saying, I come back to the same thing. Where is the data? How is it protected for when you get the accusation from the lawyer that, hey, my facial recognition template was stolen because of the association and was used to go do something bad somewhere else. So we're gonna drag you into this lawsuit.

SPEAKER_05

You know, John, I'm surprised Dawn and Greg haven't brought this up because discrimination lawsuits are the worst lawsuit you can have in our business. They're big losers, big money. That's why I made this slide. You oops, sorry. And heaven forbid that person believes it was targeted or there was some other discrimination kind of factor. Um I think you touched upon us a little bit, John, but I want to jump back to it. How accurate is this? And is I know I'm gonna you're gonna save it for the end, but is this how can boards determine how accurate it's gonna be? Because I was surprised you told me it's only 70% accurate. That makes me nervous to a certain extent.

SPEAKER_03

Yeah, so so that was it, it's not that it's 70% accurate. So if 70 they did studies back in 2024, I think it's still going on, and they were saying if 70% of your face was obscured, they were still getting 97% at uh like true positive rates. So it's it's it's super it is accurate, right? But that's all in the eye of the beholder, right? So you guys were you were talking about discrimination lawsuits, which brings up the question about bias. I am not a lawyer, I don't pretend to be one. However, I do know that, right, within this technology, there's two different things that often get conflated: facial recognition and then facial characterization. They are two separate things. And a lot of times when you hear bias, you're actually talking about facial characterization, which is identifying somebody by their race, sex, color, etc. So the common vernacular and in what I'm seeing out there, every report is that yes, facial recognition can be biased. But when you look at where that bias comes from, because you get people being like, it's a computer, how can it be biased? Like it isn't these ones and zeros, like I just said. Well, it does. However, when you start talking about artificial intelligence, neural networks, deep learning, it all depends, well, not deep learning so much, but it all depends on how like the data set that it's reading, right? And it's comparing to. So if you have a data set from an algorithm that was designed in, I don't know, Africa, that is gonna look a lot different than one that was designed in Europe. Okay, the good news is that they're understanding that and that so there is a there is this um effort to get as as a large data set as possible to train your model. And that's why I keep going back to if you're gonna pay for a software, make sure they have a very good algorithm that you're probably gonna pay for, and then that that will actually help you against some of this bias in the accuracy complaints.

SPEAKER_05

Would that be a factor though for just basic access, you know, verification? Or is that more get into more of you know, um criminal enforcement type of things?

SPEAKER_03

Sorry, raise my hand there back. Yeah, no, yeah, like I said, this is it's it's really for the characterization side, right? When you're trying where that bias really is a problem. When you're just measuring one-to-one, right? That is it's you know, verification. It's not not really an issue, or at least I've seen, I haven't seen it come up as an issue uh applied to that because you're you're comparing your face, you're looking for your face in in a database that your face has, right? And so it's looking for that, and it's very good at doing that.

SPEAKER_05

Are you if you're not comfortable, fine, but can you give me an example? Because I'm really I'm still maybe the rest of the panel's getting it. I'm still having a hard time actually understanding it a hundred percent. Can you give me an example of where um AI with facial recognition would be have a bias compared to um not making an error or something?

SPEAKER_03

So I I would go outside of kind of the law enforcement intelligence realm and look at it more from uh a commercial or a business standpoint. You have X Company and they are running cameras in their store, and they are not only mapping where people, you know, transverse in their store to understand product placement, but where people are lingering for more than three or four seconds so that that can inform them. They might also want to know the demographic of their customers. So they might try to figure out okay, how many men do we have coming in, how many women do we have coming in, how many children do we have coming in, how many uh African American, Hispanic, that sort of thing. And I think when you go down that road, I think that's where um I think that's where the models, you know, tend to uh get biased, and and that's a much harder, like I said, that's that's more of the facial characterization than it is the recognition piece.

SPEAKER_05

So, Greg, that makes you even more nervous, right?

SPEAKER_02

Yes, definitely does. I mean, it's just it could come down to another way to discriminate. I try to relate this over to associations. I mean, if they were using facial recognition in order to uh do the pool hours and keep kids out at certain hours because it's adult-only hours for the pool, which is a dangerous thing. I could see ways where this could be thought as a great tool, but misused by a board and get them into trouble with discrimination.

SPEAKER_05

Don makes you obviously you seem nervous about it since we started right away, right?

SPEAKER_00

Yeah, I I'm just I'm more about the exclusions, right? We talk about cyber liability all the time, but we are seeing court cases in connection with biometric data. And as you pointed out, discrimination, privacy, and then of course the law. And not all states have laws, only three states actually have laws pertaining to biometric data, and about 20 states have privacy laws. So the insurance companies have a battle because they may have exclusions pertaining to laws that you violate, but if there's no law yet, so we're seeing some cases get dismissed and some insurance companies having to pay out very large, even though they're excluding data breach and privacy and/or information, etc.

SPEAKER_05

All right, let's go to the lawyers here and let's just I'm gonna hit a couple topics here, and I'll start with Dave. Dave, the cops, um, they want to see it, you know. Um, maybe there was a burglary in the building and they want access to it, and then heaven forbid, they accuse the wrong person and the association gets themselves in trouble, or maybe they don't want to tell you why they want it, right? Um, what's the rule of thumb gonna be here as far as like handing? I guess this would almost be the same with video, or maybe the police, let me go this direction, Dave. The police say, hey, if you get an alert on this person, we want to know. What's your commentary on that, Dave? Any concerns?

SPEAKER_04

That's a really good question. I don't know. Um, I I don't I don't I don't I don't know what I think that I don't I don't know. I don't I frankly don't know. I don't I I mean clearly I guess if you get some kind of search warrant or judicial subpoena, I guess you're kind of stuck. But like can if the police come and ask for a copy of your clubhouse camera because it because it caught an accident like out on the main road, can you voluntarily like I don't know that there's a problem with voluntarily providing information uh or facial recognition data that you might have at your front gate to see if a person was there at the time to verify an alibi? I mean, I'm just thinking about you know law and order shows I could watch and see what comes up in these things. Like detective comes up and he's like, Hey, uh, was this person this person come into your front gate at 910 and you have facial recognition? You have no idea other other than just facial recognition. You say, I don't know, here it is. I mean, I I don't know what the liability with that. I don't know if you have to comply, I'd really have to think about it, frankly. Uh, there may be issues also on the stuff you sign with the company. The owners may have to opt into using it. I mean, there's all kinds of crazy sort of like things that uh you have to think about. I'd really have to think about it, frankly.

SPEAKER_05

All right, before I go to Sean on this, I want to expand on it. John, the technology already exists, even though it's kind of movie-like, where theoretically um the police could ask that to tap into it, right? And say, hey, these are the 10 most wanted, right? I'm exaggerating here. We want to use everyone's facial recognition systems to catch people, right? And we just we want to have access to it. And here in Atlantic City, the police actually have access to our cameras outside our building on the perimeter, and they do that up and down the boardwalk here. And and most of the buildings here agree to that. It's all outside, though. It's not inside. That is also, I mean, it's it we're already there with the technology, John, or pretty close to it. Before I go to Sean on what he thinks about the legal aspects of it is.

SPEAKER_03

Oh, we're there, right? Uh, a murder happened two doors down for me last month, and I live in a pretty nice neighborhood, right? Um, and you'll see very common people uh the officers are they're gonna go look for video 100%. Um, they're gonna look for good quality video. They want good quality video. So if you're gonna invest, invest in good quality video. Um, there are their programs where they will give you 250 bucks, 350 bucks, you know, to install the camera and allow them access to that. I I don't see I don't see them asking for your facial recognition software capabilities. I think they would have their own, right? What they might do is um exactly as you said, hey, we have a be on the lookout for this guy. We're trying to find him. Hey, if you have facial recognition software running in your in in your you know um your apartment building, can you can you just like let us know when he comes back? Like can you just like set him up as a be on the lookout for? And if he comes back, just let us know. For me, that's an incredibly valuable tool because 90% of my time looking for criminals was trying to track them down. Um, there was a case where I knew somebody was at the hotel and I had to go old school. I had to go to the front desk and tell them, hey, can you lock this person out of their door so that they have to come down to the lobby to get a new key? They did that. We saw I went back up, and the guy was like, Did you lock me out of my hotel room? I was like, Yeah. So it's just different techniques. But yeah, it's it's it's here.

SPEAKER_05

Let me let me go and Sean. Here's a here's a scenario I see happening, right? Um, you're on a board, you go to the board meeting you, I want to have this technology because we don't want criminals in our building. So I want the police to be notified. If there's someone who's been accused of a serious crime and the police have the ability to use our technology to find that person, that's gonna benefit our association because we don't want a murderer, we don't want a rapist, we don't want someone who's accused of burglary on the loose. We want to give them that information. We want to allow them to have access to it. What do you think about that, Sean? Because that's definitely it could be very much possible.

SPEAKER_01

I mean, I think that it's it's gonna need to be uh reflective of what the community wants. I mean, the you know, the association's gonna need to develop a policy uh for how it implements and uses this technology, and it's gonna have to be transparent so that the the um community members know what it's doing and what it's not. Um it you know, it has to have a you know consent element to it where they can opt out. Like, I don't want to you know get into the gate by you using my face. I just want to use my key fob olds old school, and I'm just opting out. Um and and you know, and that policy is also going to contain what are we gonna do if someone asks us to voluntarily um give the information? Is that something we're going to do, or are we gonna require a subpoena or search warrant? And and I think that that's you know, uh uh implementing a best practice on that is gonna be reflective of what the community, how far the community wants to uh to let it go. And that that get back to that scenario that you asked about, it really shouldn't be a I mean, it obviously the board's got to make a decision, but they need to make an informed decision based on what the community wants it to be.

SPEAKER_05

I'm I'm torn here. You take my building, right? If they have the technology where I don't want them to catch a guy because he has a parking, a parking ticket, but there's a guy that's a serial killer and he's living here at the building, and you know, the the the technology could give them a heads up on it. Sure, I would love it. So I guess I'm guilty also of not being sure what I really want and don't want. Um, John, there's a commercial use aspect. Back to this too. I meant people will pay associations for this information. Um, I don't know if it's accurate, it's what I found in my research. It has it's valuable. Have you seen that at all yet?

SPEAKER_03

Or you know, I haven't seen it, but absolutely the digital data that you create every single day is extremely valuable to everybody, not just businesses, but you know, um, I know that when I was working for the government, we were um we were we were told on a regular basis, hey, uh foreign foreign uh operatives and in nation states are very interested in your digital data and will probably buy for it. They want to know where you're going, they want to know your your pattern of life, so to speak, right? And um I think that the commercial industry has been way ahead of this for many years, right? Because they're using data to drive their business decisions, uh, and there's so much more data out there than there used to be. I can I can recall uh lots of instances where some of this actually had had been leaked and um information was out there as far as where people go on a daily basis.

SPEAKER_04

That that that facial recognition though is or are you talking just about like credit card numbers? Like how does how do you sell someone's face? Can you you know can is can you do that? Like can you can the company maps the face for the front entry security and then you can actually sell that to like a and then how who would buy it? What what what usefulness is that?

SPEAKER_03

Yeah, I mean that's interesting. Let me go back real quick because you you guys were talking a little bit about um uh databases and uh integrity of databases, making sure that they were they stayed out of the hands of you know malactors. What I've seen is that a lot of companies because they're concerned about the privacy implications, right? Because they don't want to get sued just like you guys don't want to get sued, they don't want their company to go down in flames because they had a breach. Um, a lot of times masking of the individual, their personal identifiable information, PII, is built into their business plan. So they will not say, um, you know, Sean, that here's what Sean's digital face print looks like, right? They'll assign him a number, right? It'll be a number, it'll be anonymized, right? And so a lot of uh companies are doing that. Uh, I don't want to say to get around the privacy implications, but to but to make sure that if their database leaked out, they wouldn't have your name, your social security number, all of that stuff, and your face print. They would have XYZ123, and here's the face print for that person, right? It's only when you start mating your enrollment in the name with that information does it get do those implications come in back. But um, where this is important is it's not just the face, but it's your phone, it's any sort of tracker that you have through GPS, it all talks to one another, right? So, yeah, you have an iPhone, you have an Android, yeah, your digital face print is on there, um, but it's also there for your banking and all of the information in your phone or your computer, like it's it's sending that data out to it to people who want to buy it, right? And it's your geolocation data, it's your spending habits, you name it, it's sending it out.

SPEAKER_05

You know, Dave, this is gonna, I'm sure this is probably already happening in places. Basically, if you have the facial recognition, it can also see what clothes you're wearing. And my building here was Memorial Day weekend, right? We're off the boardwalk here. Thousands of people came in and out of this building. So if you had facial recognition, the AI could read what kind of clothes you have, it could read what kind of jewelry you're wearing, it could also tell what kind of car you drove up in in the garage here. It's all there. The technology already exists. That's so valuable to people who want to market the people. If you could immediately know what Dave Byrne wears, what Dave, what jewelry he wears, what times does he come back and forth, what kind of car does he drive, how old is his car? You could just take this further and further, further. Maybe the tires are worn down. It'll get to the point. And Dave needs new tires down the road. Maybe your car sounds terrible and an auto repair place will contact you. The technology, am I wrong, John? The technology is already there right now to do all those things to a certain extent. I'm exaggerating a little bit, but I'm not too far off, John.

SPEAKER_03

No, you're you're not off at all.

SPEAKER_05

You're 100% so that's where the commercial use value will be. And I could see associations down the road getting in on the game and making a considerable amount of money in that regard. Um, Sean, legalities of that.

SPEAKER_01

I I gotta think that that's um something that the community would not want. Um, and I think that that's something you really have to take the temperature of the of the folks that um that live there that are stakeholders in the community and um and make that part of the policy. Or are we are we going, you know, yes, we'd like to have this money that somebody's willing to pay for all the data on our our members and all the you know people that um live in the community, but I mean, do we want to do that? And I think most people are gonna say that's a hard no. Um, you know, so yeah, I think that's got to be a decision that that um that the board needs to do what what the community uh would prefer in that regard. I think most of the time communities are not gonna want to do that.

SPEAKER_05

I was so gun-ho for this technology. Like I would, I still think I am gun-ho. If they wanted to use this technology in my building, I think I would be fine with it. I would love not to have a fob or anything, but obviously with some of the re I don't want them selling my information, though. But I'm almost reaching at the point now where they're gonna be picking up that information from me anyway when I'm walking down the street or anything. So, what's the difference gonna be? Um, vendors, Dawn. I don't want to gloss over this. Associations can get in trouble also if their vendors collect the information and they have a data break, correct?

SPEAKER_00

Absolutely. Absolutely. And um, you know, a lot of our insurance programs aren't quite up there on speed of that. So again, discrimination of the vendors and would the insurance apply or not, because a lot of our exclusions right now are pertaining to discrimination to our own or third parties. But when we have data breach and it's of a third party, how do you even control that? The notification process, it could be severely delayed. So it's so important that we do background checks on our vendors and know what their continuity process is in the event of a breach or where does their information go? How long is it retained, et cetera?

SPEAKER_05

All right. So, John, I think we dove deep. I'm gonna go to the panel, see if they have any questions for you. I want to leave you time at the end here. Um, is there any technology aspects of this? I think we've explored everything from where it is currently and where it could go in the future. And what's scary, it's already there for the future. It's just not being that widely used in that aspect, uh, or some of the aspects we brought up. Is there any technical aspects we haven't brought up today as far as how this technology can be used? I think we covered a lot.

SPEAKER_03

Yeah, we covered a lot. I I just wanna I I don't want to like I don't want to bury the lead here, right? Like facial recognition is is just the tip of the iceberg. Like I said, there are algorithms out there, and there are companies that are selling a suite, right? A suite of algorithms, and facial recognition is one, right? License plate reader is the other one, right? Like uh and I know um I know uh you know Flock has been quite controversial and has been installed in a lot of communities um throughout throughout the United States, right? Um anything from uh a person walking down the street with a gun exposed, um like I said before, somebody starting a fire. So I think that you know, if if you go out and you start looking at different vendors, I think I think the narrative is going to slowly start to change from, oh, we have this facial recognition uh technology to hey, we have this suite of AI algorithms that we can run off of your cameras. And like, what works best for you? Do you want a license plate reader on this camera? And do you want facial on that one? And do you want that on that? And so you can actually have this ecosystem of really cool things that will give you all of the information that you ever might want to need. I know that's scaring a lot of people.

SPEAKER_05

Actually, I just saw dollar signs for Greg and Dawn for some reason. This is all like stuff. People are they're gonna come up with new insurance policies. You're gonna have to buy more and more insurance for all this stuff. It's true, Dawn. Don't you think so? I mean, seriously, this is all gonna come down to insurance for the most part.

SPEAKER_00

You know, uh yes, but also it's important that we prevent the risk. Insurance is there to pay and to transfer financial risk, right? But we have a lot of other ways to transfer risk, and the most important thing is knowing what the risk is and then how to mitigate it, maybe remove our situation, prevent it, etc. So Greg and I don't want to hear about it afterwards because that's usually when the claim is occurring. And these policies are so inexpensive, everyone. $500, $1,000. We're not talking about breaking the bank here. And Sean and Dave are wonderful attorneys, but I'm sure that their hourly retainer, their retainer and their hourly services are a lot more than the premium you would pay in order to have insurance coverage for these type of claim scenarios.

SPEAKER_05

Unless you buy them lunch, then they'll work for free. They will work for food.

SPEAKER_00

Oh, I don't know about that.

SPEAKER_05

Yeah, if you take them out to the movies, if you take Dave to the if you take Dave to the ball, the um hall of fame for baseball, he'll he'll do that. He'll take that in lieu of bill blowers, and you can ask him questions.

SPEAKER_04

I would, I would. And take us to Devil Wars Prada, too. I want to see.

SPEAKER_05

Um, all right. So let me go through the panel really, really quick. Sean, any technical questions for John?

SPEAKER_01

Uh just not John, what are your thoughts uh basically on on data security and destruction policies? I mean, what what makes the most sense? What's what's a good you know industry standard for that? Yeah, you know, I've seen anything.

SPEAKER_03

I I think it depends on the industry, right? Um I've seen records of notice that you know retain information for 30 days, right? Um, I've seen things that retain it for 180 days, depending on state federal regulations. So I think you mentioned it earlier, building out policy, that's that's gonna be super important, was always something that I looked at when in when introducing new technology. Um, I think people need to take a look at it and say, for from a retention standpoint, right, you guys should already be retaining video for a certain period of time so that if there is a breach, right, you've limited your your exposure there to only 30 days. And so um that that should be considered. But I've seen it, like I said, I've seen anywhere between 180 days, depending on um like the application for uh guidelines, uh, and then uh other just internal ones of 30 days or so.

SPEAKER_05

You know, Don, you were at the last live stream we did with Tom. I think you said 30 days was they kind of felt like an industry standard or something. Don't hold me to it. Greg, any quick question? Because I want to leave John time to wrap up here.

SPEAKER_02

I just want to add one comment on of the most associations have a property manager, and we're talking about our cyber insurance, and Don mentioned how inexpensive it is and everything. And a lot of times we get back with the vendors that the property manager carries the cyber insurance coverage, and that doesn't count. The property, the information we're talking about that could be stolen is the association's property. So the association needs to carry their own cyber insurance policy to protect them. Um, my quick question for John though was and to the attorneys, I don't feel like my face is like my social security number. We're outside all the time. We're walking around, we're in plain sight, we're being videoed. How easy is this going to be for people to copy these details of my face? Which, at least if we keep it separate in the database of social security numbers and dates of births and names, is good, but it's out there as soon as I walk out my door.

SPEAKER_05

No time to go to the lawyers. John, could you just tell Greg it's super easy and there's nothing you can do about it? That's that's where we're at.

SPEAKER_03

I mean, I don't I don't want to give him a heart attack with going to YouTube and looking at some of these hackers who not only have cloned people's faces but have cloned their audio um to the point where their their mother doesn't even recognize them. With a little bit of skill and some technology, you can do pretty much anything on it. I mean, you got a point.

SPEAKER_05

Uh Dave, uh let me go to Don. Don, any quick question? No, okay. Dave, any quick question?

SPEAKER_04

Aside from whether John has an alibi for this murder that took place two doors down, no. Okay.

SPEAKER_05

All right. I want to ask Dave a really quick question. Quick answer, Dave. I'm telling you right now, like I would I'm curious what you what you would think. I I'll tell you from my building, I would still go with this technology with certain limitations. I'd be all for it. Dave Byrne, what do you think? Because I know you have strong feelings on these things. And you got to answer in 30 seconds or just pass on it.

SPEAKER_04

I I'd I'd I'd I'd be willing to roll with the with the technology and see how it can make life easier.

SPEAKER_05

Wow. Okay. All right, John, you got um you got two minutes here. What should boards be thinking about? They like this technology, it has a lot of benefits. I know we've scared everybody, that's what we do. Yeah, but boy, what are the positive? What should they be asking?

SPEAKER_03

Here it is. Oh, open up your aperture. Don't just think about facial recognition. You guys mentioned a lot about liability. I get it. But I think you can actually have some uh insurance protection if you have an algorithm running on a pool that will detect somebody falling into a pool, right? I think you can have some liability protection for uh getting notified if you have an exclusion area that you know is dangerous in your property, you don't want people going into, and you get alerted when people are in there. So I think it's both ways, right? Um, here's the thing: if you are looking at investing this, um, as always, when you're looking at technology, be careful of the technology push versus pull. If somebody comes to you and says, I have this great piece of technology, be very wary, right? That's what we call technology being pushed onto you. You want to pull tech to you, right? And that means you want to figure out your application. What do you want to do with it, right? Spend some time, talk to some people, build up requirements. And then when you do start, when you do your research and you go out and you start looking at vendors, you have a list to tell them and say, hey, this is what I want done, or you don't tell them, you just say, hey, like what can you do? And you're like, nope, nope, nope, yes, right. Um, that that kind of protects you again against going down this rabbit hole of the latest and greatest technology. I already told you that the algorithm is the most important. So if you get into negotiations with a company who has a nice software, you really need to understand where their core algorithm comes from and uh and who developed it. And then um, I think it was Sean that mentioned this policy development. You have to have policy. Very often we will bring a new technology to to the game, and then we'll be like, oh my God, I didn't think about that. We need to narrate some policy around it. Nope. I think I think a reasonable approach to be is hey, we're looking at this, let's start to build some policy around it right now, right? As we're going into and as we're looking at technology, because that can help us in inform that policy. Establish your requirements, take a look at your uh the algorithms, and then not only that, uh I said the cameras, the cameras are also important. How detailed they can get. So if you're looking at um, you know, infrastructure investments, uh buying new cameras, you're gonna want to make sure that those algorithms can support that uh to set you up for success.

SPEAKER_05

John, I meant thank you so much. I I learned a lot. I'm equally optimistic and excited and scared at the same time. So that's that's how I feel right now. But thank you so much for being here. Much appreciated. Thank you so much for panel for being here. And I hope to see everybody soon. Thanks, everybody. Bye.

SPEAKER_04

Oh, it's a pleasure. Bye.