How Behavioral AI Helps Fight Cheating

[music] Welcome to the Navit Gaming Podcast, where it is our mission to explore the business and future of video games. We bring together the industry's brightest builders, investors, and thinkers to keep a pulse on current events, dissect emerging trends and games, share lessons learned, and have a great time. This podcast is also part of Navik's growing ecosystem which ranges from free and premium research to consulting and advisory services. For more information, visit www.navic.co. This episode is brought to you by metal.tv. Built by gamers for gamers and used by over 15 million gamers per month. Metal is the fastest way to clip, save, and share gameplay on PC. Millions of clips get shared every week, turning your game's moments into word of mouth at internet scale. With Discovery being one of the biggest challenges in the games

industry today, Metal is building a platform where publishers, studios, marketers, or anyone looking to reach gamers can find the right players across multiple genres and geographies. For more information, head to metal.tv or check out the link in the show notes. This episode is brought to you by Heroic Labs. Whatwap, a studio known for its viral card games, thought matchmaking couldn't get hyperpersonalized in real time. Heroic Labs showed them otherwise, moving from elemental to sophisticated, economyaware matchmaking. This fundamental decision gave WWAP better balanced matches and steadier handling of traffic spikes without the system breaking a sweat. It's a small change that reshaped how their whole team thinks about matchmaking and multiplayer. That's the kind of shift Heroic Labs makes possible. Not just faster infrastructure, but better decisions about how your game should work. Curious what metrics changed and

why? Find the whatwap case study on heroiclabs blog by going to heroiclabs.com or simply check out the link in the show notes. Now, let's jump into today's episode. Hello everyone. I'm your host Devon Becker and today I'm delighted to be joined by Andre Pimementa Rivero, CEO and co-founder of AnyBrain. Anyrain is an AI anti-che platform that detects cheating by analyzing how players use their mouse, keyboard, controller, or touchscreen rather than just scanning game software. So today we're actually going to discuss how AI is changing both sides of the fight against cheating. Both for the cheaters and for the anti-che. Hopefully more so for the latter, but we'll see. Uh but just to start things off, Andre, can you give us a quick overview of your background and uh how research on mental fatigue actually turned into an anti-che company?
Yeah, sure. Well, first of all, thank you very much for the the invitation. So, I founded any brain almost 11 years ago. Um I started doing um a research on University of M. It's local university

on looking for human computer interaction and how um we need to use some of our cognitive skills to um to use machines on the time uh we were developing an algorithm to detect mental fatigue and how we started any brain. So I was doing my PhD on creating an algorithm that was able to detect how tired we we are while using the computer during the time and uh we also learned that okay uh it's when we are using the computer uh it's not only about being tired or how we perform it we can use this information for more stuff. uh but we decided to start the company then uh we initially tried to explore the mental fatigue side uh on the beginning not only on the gaming we work with different different um industries uh where people were spending time on the computer at the end what we wanted is to give some advice like stop for five minutes or come back in a in a few we were introduced to the gaming when we

started working with esports organizations in uh 2019 um selling same software We had uh some requests on some performance coach. They were asking hey can you apply your tech for esports because it's hard to control the the players or helping them to perform better. Uh we said yes and that was the moment that okay we need to do this uh for gaming industry and not only about mental fatigue. Uh this can also be related uh with cheating about identity because at the end one thing that in the gaming it's unique we need to interact with the games to be able to play. uh they were made on that way. Uh so uh let's export human computer interaction field with the that was uh our uh expertise.
Interesting. So definitely like on the behavioral side of things looking at that analyzing it and then figuring out hey this actually helps us figure out if people are doing things maybe that uh they normally wouldn't be doing that maybe the computer's doing for them or helping them. Uh so out of curiosity then how does detecting cheats from like player inputs work and how is that

different from like a traditional anti-che?
Yeah. Uh so the the main difference is that we look for uh the behavior of the player and not for the machine itself. We are not looking for what kind of software you have installed. If there is something that is trying to access the the memory of the game or trying to tempering with the files. What we look is okay if someone is able to do it because we know at some point uh they will we will understand if that behavior is possible or not. um and how this beaver can translate in detecting these some kind of hacks let's say. So as I started presenting uh the human performance while using the computer is unique per player and uh it's also unique between humans and nonhumans. when when I'm talking about non humans could be bots could be a script for an aimbot could be an advanced attack that is helping to aim is helping to adjust uh for example the recoil or is uh taking decisions as we are seeing nowadays a lot of AI bots as well that

can take decisions they can uh move on uh levels more than a static uh software so with this uh we look okay uh let's I will give the example of an aimbot that Uh it's probably one of most known hacks on on gaming mainly on the FPS. Uh we need to aim on opponent and uh the way we do it if we do it faster will be better for us because it will give us an advantage and here we usually are talking about seconds or sometimes milliseconds of reaction time can can do a difference on on my performance. So the when we look for pixel by pixel and time time stamp uh I mentioned pixel by pixel because we analyze all interactions that are performed with the with the mouse we know okay that sequence of actions that sequence of times or trajectory is not possible for a human uh that is giving a small advantage. There was some adjustment and the idea is that okay we will understand uh how this input is helping the players. How going back to your question

how we are um creating a different approach from the traditional solutions it's again because we don't look for the computers or software itself and nowadays uh without the DMA second PC uh it's moving a lot for input uh so most of us are trying to mess less with the game files also because they are more protected also because the other solutions they also doing a good job um or protecting the the game operating systems they are getting more closer as well. Uh so at the end what we are interested is okay if you bypass uh game defense because you were able to do a good reverse engineering to access the memory or because you are a second PC we don't care what is your setup uh we know that it will change your behavior. So you mentioned that behavior is personal, right? Which I mean sounds obvious but like that matters a lot for for behavioral antiche. Is there some level of like training off like standard player behavior to understand how the player nor performs? Like let's say you

know I first start playing and I'm just insane performance at the game. Is that then treating that as my baseline? And then does it have to then you know kind of go well it when when does your baseline start and when are you kind of out of bounds? Whereas like if I'm cheating right away then my baseline's going to look like I'm just really good at the game.
Okay, that is definitely a good question because a lot of players they mentioned that that could generate false positives on this approach and they are less concerned with that. So uh let's say that uh you today you release a new game, no one has experience on that game. There's a unique mechanics on that game. So definitely your gameplay will change as you learn the game and uh as al uh release new new updates. The way we handle this device is already present. So it's unique the angles how we we manage that what we do is we process information on the screen and depends how it is being presented by by the game and also what kind of dynamics the game has for us to interact with that game. But that will never change

the way we are handled this devices because we are using the the computer or the console or the the phone for a while and we already have this pattern on us. And also there is a limit between uh what the human and non human can perform. Again example is some movements they go so fast on the direction of the opponent getting an easy example that it will be impossible for someone do it or without big movements for example. Um so uh that is the way that it's totally agnostic to the the game as we don't for example we don't look for your your stats in game. Uh now for example if you are doing more head shot you are getting a better uh kill assisted ratio. Uh we are not looking for that. So you can improve a lot on the game. Uh that it's totally independent on that. There is also uh on our approach two different ways. So uh I didn't mention before but we also look on identity uh for boosting and smurfing. So uh on this we create a

biometric prof profile for each player uh that we learn how you handle the device um and angles movements and each g and each person is unique the way we we handle these devices. Uh it's proofed and it by the way it was used a long time ago as well on banking for passwords. the way each person typed the password it's unique. So there is some start of state of art on that. Uh I'm playing games is also unique. So that profile is created just for identity. And then we have models that understand what is possible in terms of uh movements handling a mouse or how you handle a controller or how you do your touch and that once they verify if that sequence of actions are possible for humans or not without needing to learn from you. Uh, it's more about how people interacts with the device.
That makes sense. I mean, I'm I'm curious about something you said. You mentioned smurfing. Is this a tool that can also identify smurfs? Like if a game, for example, wants to punish or

wants to adjust the ranking or things like that.
Yes, exactly. Uh, we have some customers using uh these features. So, different usage uh when someone is just trying to uh to mess with the rankings to take advantage. Uh also when players uh are banned and they come with second account IDs also to help to stop them from the beginning before they do uh damage on game as well. Uh again because most of them some of them they stop when they are banned but most of them they just create a second account and they try to adjust their hacks or whatever and try to do the same. So yes, that are the probably two scenarios that we help and also on the well boosting because it's kind of a reverse for us is connected is the same and again uh a lot of people they also try to get high rank uh without deserving to be there using this boosting service. So for us it's always connected boosting and smurfing. Interesting. Yeah, it's definitely something that I feel like doesn't get talked about a lot anti-che. So that's why I find it interesting that is a

solution to help with some of that especially around just matchmaking being complicated these days. Um, now a lot of games are are stacking like multiple anti-cheets together both like at the software level looking at a lot of the stuff you mentioned around u like actual software hacks and memory access things like that. Um, where does the behavioral detection sort of fit in that stack? Are you seeing like it being used with other like kernel level stuff and pieces of software that detect the other things or is it more just like this is enough by itself? Like where do you see that fitting in?
Yeah. Uh, definitely not. So we believe on a stack of security solutions to protect the game. So we need first start protecting the the client itself uh the game binaries uh encrypt the positions of memory ofation everything that will make the life harder for for shooters uh and also to try hiden information because at the end uh what gives advantage is to collect that information. could be a position in memory that says where is the opponent uh um position that then I can use that

to take advantage or even in some case modify the game. Uh I can simply change a variable saying my health is now on 100% again and I will never be uh killed by someone. So that is I would say it's mandatory and you need to use that. uh but and has as I mentioned in the beginning it's not enough because we have uh this fighted score on colonel level who first starts the machine if he's the cheater or even the car colonel level at the uh but also because most of these axi are moving for uh outside of the machine itself. So you could start an kernel level uh on the machine itself but most of hacks they are getting access to hardware. So you you transmit or your video uh for a second PC uh so both PCs are sharing the the same image. Uh you can inject input uh on a mouse controller or simply with a remote play. I would say that one of

ways uh that easily to hack consoles is to use remote play because okay consoles they are very close to environment uh which is good because le hackers have access to to to that and uh you can't run your own software so they are much more close to the environment than for example Windows or even Linux uh but uh there is a lot of new hacks that are being developed for consoles because the the point was okay. No one thinks it's possible to hack on consoles, but it is. Uh there is hardware for that. So, let's build a lot of hacks and move to there because there is also less solutions on on in place that will make uh uh safer for shiers let's say. Uh so again uh at the end uh we need to protect and there is good solutions there also lot of efforts uh on this area and I'm happy to to to to hear and to follow the progress. Yes. Also in terms of the the operating system, we are seeing Windows releasing

features that anti- sheet teams and solutions are adopting like uh TPM um other solutions that will uh make the the operating system more closed. Um and I believe at some point we will uh play games on Windows like we play on consoles with a special environment that will be very closed to gamers and if we want to play we play on that environment. If we need to work and uh for we have a different environment that will be more open and I guess uh that will be okay because uh what we are saying is while you play you have this closed environment. Uh but that will just reduce the the surface of attack because then we have all the stuff that is already possible with remote play and other stuff that we can uh we can perform without messing with the the the operating system or the miss the device where the game is running.
Nice. something you mentioned that made me think of the the problem with keyboard and mouse uh inputs on consoles which generally is usually not allowed

as as part of it but then you know there's the devices and things like that. Um, you know, first question is like is that something that you're able to detect because it does seem like something that would be in your wheelhouse to detect even if it was obfuscated in some way like remote play as you mentioned and second thing uh related to that is is actually keyboard and mouse input on uh mobile games which is often done through like emulators and things like that. Is that also something that you're able to detect? And I mean do you work with mobile games or is it mostly just console and PC?
Yeah. Uh no, we also work with mobile. Uh we'll start first with console. So the main point of using a mouse and uh more than keyboard is to get access to the aim assist that is supposed to be only um well it's active for controllers uh because it's more difficult to aim and handle for most of the people. So the games they develop this mechanism that helps to aim while you play with the controller. So some people just get advantage on connecting a mouse. uh they um they fake uh the input saying that it's

come from control but at the end is the mouse and that creates a problem because gives you really uh good advantage on that. Um we can detect that uh that was probably the most common hack on consoles. Uh but it it is changing. Uh it's not anymore just the mouse and keyboard emulation is starting being about and no records. Uh some controllers they already allow you to record movements as well as we have on the with the m most of the mouse on the the PC with all of the offer to make them different that can record or you can uh build your own macros. Some control devices are already allowing that but also because again with remote play and not need for remote play. I will not name the the hardware but some some hardwares are possible to connect to your console and um that will capture the same uh output video output that is being sent to the console to a second PC and then again you can inject on the

computer some movements that will give you an aimbot and uh these aimbots and noro they have more impact than uh the mouse and keyboard emulation. Well, that is already good because usually uh when you develop these hacks they are uh very efficient. Uh so it's starting um having more and more hacks on these environments. Uh going for your question about emulators. Yes, we also have we we work with some mobile uh games as well. Detection of emulators because usually emulators uh can represent that you are playing uh well could be legit. Uh there is a few case where you are just playing with emulator because it's easier you have on your computer. uh depends on the game. Some games allow, others not, but most of the time it represents that you have some software that is more easy to run on the computer than on the mobile uh itself. Nowadays, more and more games are are being played on um on the PC because the

integration uh Microsoft Windows now has a way that you can play games. Amazon was doing the same. Uh so some games allow it. Uh usually it's by uh decision of the game. Okay, I want my game to be played because what I want at the end is more players. So more platforms better. But other games they have the uh competitive part and usually is where the concern about anti-itis uh for the mobile while on PC and consoles we have all the competitive environment could be because it's a multiplayer and if you are killed by someone with an inbot will feel okay it's not fair uh but other games it's about leaderboards and most of mobile games you don't compete in real time if you yes but still you have the leaderboards uh you have your own account with some value that could be destroyed by a bot that is farming and then uh your assets has less value because of that. So emulators are one of

the problems uh connecting mouse and keyboard for mobile game I don't see as today a problem um because what the games on mobile they don't develop special mechanisms yet for mouse and keyboard if you play with that. Uh so uh it's the same you simply are running it on another machine but again when you run emulators because you have a full setup on your computer usually that gives you uh more tools and more mechanisms than a normal player.
Makes sense. Definitely something that as you said is trending up and also like the Google Play stuff on PC and just continuing to increase like that sort of crossplay.
I imagine just you just have to do it where it where it makes sense for the developer then I guess. Uh when should a studio start thinking about anti-che? Like when when do they need to start considering that? Is this like something they need to be thinking about very early in production? Is it more like okay we're going to go live? We need to start integrating. Where do they where do they start to plan for this work?
Yeah, I would say uh from day zero

because it's also about game design. A lot of stuff can be uh solved by the way you uh design or implement your game. I can start make sure that uh you are server authority. So stuff that needs to be verified on server instead of be on the client side. If you can have an architecture that have that in mind you will solve a lot of problems. Um going behind that part usually being from health and beta test there there is already already It's good that you can already look on that stage because uh you will learn okay on my game it's there is more norles because the the way the the guns they behave they an oracle could be very powerful um and also will give you an an view on how uh your game can be hacked or not. So and then of course as as you release the game you need to be every day uh paying attention on what is happening and trying to improve uh what is not yet

solved. So I would say that depends on type of the studio because some of them they have low budget others have a big budget but uh there is two things that I I would say that are mandatory one is about the game design architecture. So try to solve most of problems with your game engine team as well. So even having people thinking from the beginning how we can if in the future someone want want to change some items on the client we'll make sure that we protected on server side that kind of decisions and then it's about observability because even if you don't have a budget to have the best solution from uh day zero day one you will know what is happening and when you decide or you have more budget or you decide to put uh in place more uh more tool and solutions will understand okay what is the fact of what I'm doing how I can solve this part of the problem. So obser observability will be again uh about architecture that you find for your game. What kind of information do you collect? Uh of course stats about the players who is doing

crazy stats that kind of stuff can help you to understand uh how big is the problem but also when you start implementing solutions uh if they are solving part of the problem. So yes I would say from the beginning.
Okay. Well, I mean obviously like that's easier said than done when it comes to uh planning on a game, right? But like good advice to think about it from the game design level. I think that's a a smart angle to go what can we do to sort of minimize it from the design. Um when it comes to that like in terms of like the impact on games, uh obviously you've seen like some you know the the actual cheating impact when you're able to detect that you get an idea what these players are doing. What is what is the impact you're seeing? Like how does it show up in retention revenue? uh like like look where's this impact like obviously people are spending money to protect against cheaters obviously people don't like cheaters what's the actual impact on the game
yeah so um first is about the reput uh reputation of the game uh because what we see is okay I'm playing this game but

I heard it's full of shooters I will probably not uh jump on that game or will stop playing when I have uh someone that is a shitterter against me because I I will feel it's not fair uh and that affects affects a lot on the reputation but also on the exition of new players on both sides. uh I would say that um the most and and then there is depends on type of game some of them could affect the economy let's say uh especially on MMOs or if there is some component on economy uh some farming house they will be farming source and then they sell to the players and me as a publisher I will not be able to sell my package because they are buying with a low price so that affects directly the economization of the game and on potentation uh it's more indirect but you can measure okay how what is my retention how new players I'm acquiring because of this problem and I would say that uh about the reputation uh it's one of most important things on the antich side as well because if you don't show

that uh you are protecting or doing a force you have a good communication around the problem you could have the scenario where your players just feel there is a on the other side because it's a good player but at the end it's just another player here. So you have a problem without having the problem or the problem start looks bigger than it is in fact. Uh and again that will affect the perception of the players. But there is studies we also um shared recently that if most of players if they feel there is a on the other side uh they will probably try to stop the game or pulse for a while because it's very frustrating uh playing against cheaters. Well, at this point, I I'd say cheating is kind of a professional business. Like, there's a whole amount of money involved. Uh, both from the game developers you're talking about, like, you know, with the reputation and trying to keep players around and things like that. Obviously, people don't want to spend money on a game if they're not having fun, but then there's the the people that are spending

money to have fun via cheating. Uh, what does the sort of at least from your perspective, the the cheat as a service economy look like today? Obviously, we had the the older days of buying cheats, but now maybe there's like subscriptions and other things. Like, what does it look like now?
I [clears throat] would say crazy. There is that is pointing for between uh $5 billion up to $8 billion on uh economy from
Wow.
Yeah, there is some sheets you can buy and most of them they you pay for $20 uh day or a week depends on the game. And it's crazy. some do you have free freetoplay games where you are paying that amount uh per week or per day just to to be able to uh to play the game. Then there is all of these new hardwares that are expensive that you need to buy to play uh that goes from $100 to 1,000 and depends on the game and also the setup you need to build. Even if it's just a second PC, you need to have a second PC, right, to to hack the game

and that has a cost. Um so yes it's then there is all the boosting service that are around uh farming house where they again uh they need to invest a lot to have device and computing power to to farming so it's growing a lot and again what for me it's more uh it's about psychology but it's someone that doesn't pay for the game but invest al on on these hacks
uh to be able to play the game uh it's it's I feel that I'm interested sometimes because of reputation other times because they have I mean it's not not I would like one day to to see a study about the what what is the main motivations of a of course I understand sometimes about reputation others is just to try uh do bad stuff from the game for reputation uh not as a player but as a hacker or a shooter but again there is all of this side about economics on people that are

developing and making it easier for for shooters. There's also other side by the way interesting that is with the AI um it's nowadays I would say everyone is capable of programming so most of people is also able to develop some hacks I would say not probably not most of us but uh simple no recoil or uh other stuff they can build it uh without need to pay or have too too much knowledge as well
I mean that leads into my next question then which is like obviously you know you've spoken a little bit about how AI is helping on the protection side but how is AI and then as you mentioned some of the hardware cheats which may or may not be related to AI changing the actual problem like you mentioned a little bit of it earlier on but like where is this kind of going right now?
Yeah, I would say the main impact or the the first impact is on making easy to develop hacks. Uh so again I can create my own script. I can use my hardware that allows me to load uh or software. I

mean there is different options I can use to to just build macros nor aots whatever that runs on software level that uh without knowing how to program I can have that once uh this means that also makes each probably hack or unique as it was developed by me for my own style but there is also the other side that is about the capabilities so when I'm here talking about computer vision uh about AI um lots that are also very focus on uh visuals that will will help you to have a good model that in real time can give you the position of the enemy without accessing the game files. That is the point even with a camera. There is already some guys trying that with external camera looking for your screen it can gives you the enemy position better than a human and that with that you can get an advantage. So it's about processing uh pixel bots. So usual pixel bots are

an easy hack um to explain because basically it's looking on your screen when there is a pixel with a specific color that usually is used by the the enemy and what with a simple software you can just uh run what is on your screen. what are the pixels that are being presented and when you find the color that represents enemy you already know the position so you can order uh the position some some games they try to mask that randomize well uh there is different techniques to try to reduce the problem and then there is this second layer okay maybe I don't need the pixel I can process the screen or I can have an external camera that is able to uh analyze objects track them and give me that information so it's also about the capacities that AI is also in terms of a mix between software and hardware uh to allow uh to to to be harder to detect.
So I mean speaking of detection and and obviously this is like a theoretical

question but do you think at least for a temporary period that AI could get to the point where it could convincingly play like a human whether that be trained off a human as you're already kind of doing with AI to learn how the the people behave but could it then emulate how they play just better for example? Yes. So nowadays we already I would say if you if you went back like two three years most of hacks they were kind of difficult to see but in terms of inputs I would call it stupid uh kind of uh always having the same uh uh click duration same static uh data. Now players they are randomizing they are trying to use some of techniques to make it more human in terms of cures like uh in terms of uh it actually looks like more human and again these llms uh also helps on that because gives you the knowledge okay how I can try to make my input look more human uh when while I'm

developing my own hack or uh how I can try to be undetected there is already and some people were trying to field uh some some Asians to play like humans. They were learning how this human or this person is handling the mouse. I will try to do the same and the play the game and the result show that it was very difficult for anti-hit side because usually they are at least at this point very optimized to get results. So if you need to aim an opponent and you try to also uh do has the human, you lose some of the the momentum or you lose some of the extra ability that uh AI is giving to you. We don't know I would say at this point uh it's not possible but I can say in two or five years said whatever we need to to be ready for that moment. I also believe that one of the main threats uh not only on gaming but in general will be about how these bots will be good and how we will make sure that is a human or a bot uh accessing

an uh banking account accessing my website that I don't want bots there they will try to make them more human and more human uh I believe that we should be able to develop techniques and improve the current ones to detect that ones uh that at some point they will uh advantage that uh using these scripts or LLMs. Uh if you remove that advantage, behavior will be more human and uh it's it is not so impactful in the game. But I believe that we it will grow on both sides. But we can win the the this battle keeping uh and having already the mindset that they will evolve and they will try to be more human and more.
We're getting ahead of that. There's obviously like as it starts to trend that direction, there's much more of a risk of false positives, right? where the bot the more the bots act like humans, the more the humans look like bots. And we already have this problem right now with uh with one of sort of the witch hunts in AI where where because AI was trained off of humans, AI art tends to look like human art and therefore like you know we tend to

mistake both ways. Where do you see anti-che ending up hurting legitimate players both kind of in the past but also in the present where things are going with AI on both sides? Yeah, I would say that's probably one of most important topics when we talk about AI systems for for uh detection and yes there is false positives even if it's one in 1 million or whatever could be uh when innocent right so here uh for me the most important thing is to have a mechanism uh to review cases uh they could happen I would say it is impossible but games and studios they need and also work on antachy teams uh create ways to make sure that it's possible to review uh the case and refer to decision if it's a bad decision. Uh of course we should try reduce numbers of false positives and there is already as in a lot of system not only on on any brand but other people uh try to deploy systems that have a very low false positive rate but still they could happen. So this the next step and

something that we are working is about explanability. Uh because also when a is uh detected and banned they will say oh no I was playing normal something happening on my computer can you review and if there is no good tools to to to analyze and say okay he's lying or no it's true uh it will be difficult right uh and we'll probably be banning um innocent people. So it's about creating tools for uh the human uh be able to review and analyze the analyze the the case in order to revert if there is a positive or to make sure if he's a shooter uh he's still bot because he's just trying to lie uh to recover the account. So it will be a lot about making these algorithms less blackbox and give more evidence for people that uh is working on managing the communities and uh taking decisions.
Yeah, definitely. I mean obviously difficult because because everyone's

going to deny doing it, right? No one I mean it's pretty rare someone admits it until they're finally finally proven and they're like okay yes I I won't do it anymore, etc. Uh, when a false positive does come up though, I'm curious where the ownership and of that ends up landing because obviously the game makers aren't necessarily making the anti-cheats most of the time, but you're also not necessarily in a position where you could always like have full ownership for like what cheating looks like in a given game. Who ends up owning that? Is it is it the the vendor? Is it the game developer? Is it a bit of both?
Yeah, I would say both. Uh so for example in case of any brain we don't b people directly but we give the information for the studio to take that decision. So if we give them bad information it will be also our our responsibility and we need to find ways to solve the problem or to again to give tools for for them to analyze. So uh there is some uh solutions that also they do without the game but at the end uh when we play a game we accept the terms of the game right. So that is interface but the game studios when they

work work with vendors or internal teams they they trust on the the solutions that they are using right so I would say both uh both have the responsibility here making sure that the environment is safe and there is mechanisms if there is a problem uh that problem can be solved and revert
you mentioned you know with that sort of thing where when it is you know you're pretty confident that someone was cheating making it a little more transparent being like here's the proof here's how we know they were cheating so that there's less of that I'm being falsely accused. This is, you know, ridiculous. And you get the social media uproar potentially or the other way on them. Um, when you're being that transparent, isn't there the risk that you're basically showing people how to cheat in the future because you're going like, "Here's how we detect it. Here's how you know." Like, you know, people that read, you know, crime novels, how to commit crimes better kind of thing.
Yeah. I would this transparency needs to be uh of well 100% for the studio and publisher that owns the game and I would say the

maximum uh to the player but not 100% because of that and also all the GDPR and all the rules they have exceptions for security because if you share everything they will use that information to build next to one and build again and tell a point that are not anymore detected. So that's why all of the the privacy uh legislation has exceptions for uh the the security fraud side because we need private golf course even with that um in my personal opinion most of they try to be transparent as at at some point that we are not giving advantage uh but not just yes there is a solution I don't know what he's doing if he's accessing my files or not because I don't want to share with that if he's accessing files okay just share with that uh because at You also it is important to also have the trust of the players when when I mention about the on the beginning that sometimes it could be playing with other person and I I got shot by someone I was in the shiator but

because I don't trust on the the game or the environment is is not so uh so trustable and being clear on what what we are collecting what we are doing even without showing details it's a part of the gaining the trust of the players uh so Yeah, there is this barrier. But I guess we I believe we should try to be as much as we can,
right? Especially if like you know they're just accelerating what they're doing on the sheet side anyways regardless of you kind of give I mean obviously they made more sense when it was like very very technical. Now when it's like AI it's a little more behavioral and like unique to that individual not necessarily like hey if anyone you know anyone could get around it by just doing this. It's like that individual person maybe could but it's too late for that individual person I guess right
but uh what do you what do you think I mean obviously this is uh maybe maybe even too long of horizon but what do you think anti-che looks like in the next three to five years maybe that's like I said maybe too long just because of how fast everything's moving now obviously we talked a little bit about where

cheating is going but where do where do you think the anti-che is going
yeah so uh I will say that AI is definitely part of the solution and um we are seeing more and more uh different approach using AI because as it is uh growing or improving so fast with AI definitely AI is also the the solution to to fight fight it. uh personally also believe that hardwares and uh inputs it will be less about uh how you reverse or you attack your game while it is needed and has MSU uh early I believe that we'll have soon a version of uh Windows that will be kind of consoles or maybe Linux also right and just a note here that most of games they are not compatible with Linux because of it's so open I mean there is different versions of the kernel. You can develop your own kernel version that uh games are afraid on deploying there. And we have now uh

Valve uh deploying with Steam OS based on Linux will be also a challenge uh because of that. Uh but again I believe most will move for hardware and inputs because they will be external and uh will be more safe or easier for most of the current detections. I mean, you bring up a good good point here about the the link side of things, especially with Steam OS and the Steam machine. Um, obviously like VR could have cheating in the future. We don't have like a lot of that happening, but you know, it's a possibility, right? That there continues to be new devices pushed, including, you know, ones recently announced by both Meta and then the Valve frame as well. Uh, obviously the Linux side, as you mentioned, difficult to secure. I mean, to be fair, Android's not necessarily a ton easier just because it's also pretty open OS. Um, do you see then your solution as being more behavioral a little more able to deal with that sort of thing than one that is just like more looking at the the the software and

hardware level? Obviously, if people are moving off of the the software level anyways to do things like second computer stuff. Obviously, people are it's kind of weird to try and do like a second computer to try and cheat on a Steam Deck, but it's possible, right? And uh is that something where then like maybe this this behavioral style of doing things or statistical or whatever it might be has some advantage might makes it more possible to do.
Yes and no. So uh yes and we are already uh working with games that runs on Linux. Um and no on the terms on the spec that if some stuff is not protected on client side at least for uh for stuff will be made on client side and it is already happened on on the gap uh could be if for I will give uh one of the hardest scenario if you change colors on the game for example twilight an opponent that gives you some kind of wak even if it's not true wak it's already done uh on that side and you are giving extra central information to the player and

he's taking advantage of that because the game was not so well protected. Uh well or we can have again scenarios like remote play or cloud gaming where everything runs on server side or the custom environment even on Linux that will be possible and maybe we can have a special virtual machine uh as we have Proton and uh that emulates for Linux that will be like that. So uh that's why yes and no we are well helping on that but there is still stuff that needs to be done to try to make sure that at least the the kernels where we run the games are are loss or validated as possible.
Yeah good point about the uh the cloud gaming side as well because I do uh recall that was one of the nice things about when some of the more competitive games started showing up on cloud computing. Obviously you had the latency being kind of an issue but you did know that you probably weren't playing against cheaters and that was kind of interesting. Obviously like there is still the potential right for there to be like the the client side like pixel hunting kind of cheating where it's like

controlling the inputs and stuff like that but at least the software side is like a little more secured. Um so it's an interesting way to look at it. Uh do you see like with the behavioral side of things which is I'm sure going to continue to grow just because like this the software and hardware side is limited in what it can accomplish at some point as well. Uh do you see that ending up being used beyond just straight cheating like for things like account sharing you mentioned boosting being one aspect or smurfing uh or even like age checks which is becoming sort of like a behavioral intelligence things like even what you know Discord was announcing things like that where it's like we're using this behavioral data in different ways for security that aren't just straight about cheating.
Yeah. Even the shooting on the cloud gaming uh let's say it solve a lot of problems and I say one of the most difficult that is wak or ESP because you uh you don't have access to the game it's on the well on second machine but with that you don't have also a solution that is on the PC looking for what is running and again has we mentioned

computer vision and other stuff and then you can inject Ali boot without verifying what is on the client side so that also opens doors for for that and while on At the beginning it was for most of people a way to solve the problem because wak and others were not anymore possible but some of the client with all of these new hacks based on input AI computer vision now it's safer for shooters because you don't have even anything that is looking okay I have here this software that is well known for injecting input no I don't have access to that because uh the antach side is not running as this is this disconnection uh regarding usage of the service for uh boosting uh I at the end it will be the the same uh the the the thread is same and my point is that behavior will be more and more important here uh because it will be also one of the unique or possible angles of detection on these platforms.

Awesome. Well, I think it's going to be an interesting future regardless of the the sort of cat-and- mouse game that never seems to end. So, I'm looking forward to seeing where it goes. Hopefully, you know, less on the false positive side as a as that's never a fun aspect of it because no one really wants anti-che to have to exist,
but it's kind of a necessary evil at this point, which hopefully keep keep under control. But let's uh we just want to end with a with a few uh fun questions here. Uh just a little more opinionated. Uh, and we just started with what is the most creative cheat that you've seen so far?
Well, there is uh um a guy that build um kind of a table that was moving the the mouse. Uh so the hack was just moving mouse physically with kind of a robotic arm. So there is nothing modifying or tempering with the game and the guy was using the in this case it was a computer vision. So he knew the position and he was trying to move the mouse uh physically. So there is nothing that he

changed it but gear was created by robots. So it's kind of the the way I see maybe in the future people will have robots or robotic arms that will be holding the mouse to be play better than them. So yes that is one of the craziest way that I saw uh recently
you look for people with like really high DPIs on their mouse. I guess at that point to move much.
Um, in your opinion, and obviously like this is subjective, it's like which game do you think handles cheating the best right now that you've seen regardless of like whether using your software or not?
Yeah. Uh, this is a hard question. I don't want to mention one, but I can mention a few that are uh doing good stuff in terms of communication. So uh you'll see at this point most of triple AC they are moving on that direction and also new games that is sharing uh what is number of bands what what main threats they have um because communication is key nowadays uh because you need to to to win the trust of the

players uh when you release a new game if if you don't have a good communication about what are main threats how you are taking them uh people will think yeah this full of shers uh if there's a new game you don't have in even rankings or whatever. So you probably be playing against people that is already good and others not. So uh the difference between the the players will be uh big part of the perception. So most of the solutions they are now uh and studies they are doing more communication and making it more important. Well, let's say that probably six five years ago it was all the Adachi teams they were trying to convincing the the game design team you need to do more effort some gaming because that will affect our success and no we just don't have a good game play that will be a secondary part and now basically is changing but I would say that there is I can I can put some names um right is doing a big effort on communication but not only AI um is also

doing sharing what is type of puns So for example, Respawn recently shared that they they banned most of top players on consoles with new stuff. So this uh gives also confidence. It's not a problem admitting there is problems, right? Because the players they know it. Uh it's more about how you show to them that you are solving the problems.
Uh on the flip side that this is kind of a tricky question. Do you think there's any kind of cheating that's happening? uh you know obviously it's a big scale that you think is actually fine or like reasonable enough like maybe maybe in PvE games. I'm not sure.
Yeah, I would say probably on PVE games if you have some macros that could help you. Well, just uh not spending time on something that you can do uh or single player whatever. If you do some uh stuff that if there is no leaderboards, competitive games, if it's just for you, that will be the story. Uh I would say that I would not accept that at all. But I I could understand if it's a single

player PVA or if there is no rankings with no component where even if you have a simple macro that helps helps you to loot will not have any impact on the others. Okay? Because it's well you are just automatizing your uh your time and not affecting the time of the other players. Especially in all these like really tedious survival crafting games and stuff, right? That just take
Yeah, exactly.
I mean, obviously Minecraft bots are a big thing in AI uh testing all the time, right? Where it's like a very common thing to have AI test itself in Minecraft. So, I guess technically fine there. What will be well it will be nice for them to be um be possible to track them be identifiable and I guess in the future we also have some some fun playing against these AI bots if they play with fair information let's say if they are taking their own decisions if they are reading the screen as we are and we know there is a bot there will be nice also to see how we are competitive against some of these uh bots they are

built with AI in the future but again they need to be identified and uh you need to know It's not kind of now it's a human now at night it's the bot and then I return that's not fair
but if you identify okay why we can build some ecosystem around that as well I guess it's like the end of the day where you know there was like chess matches against computers and things like that like there's a lot of that uh involved in like RTS games or mobile games where there's AI playing against it but that's clearly designated I think I guess to your point then it's like as long as it's like well established this is AI there's no like confusion about that uh it's a little or okay I guess which has been a big issue in AI in general which is transparency around what's happening uh so cool but I want to say thanks for for coming on today uh a lot of interesting stuff happening in the antiche world and AI is definitely making it very interesting for both sides so it's good to get your perspective on the side of things uh where you're trying to use it to help uh make gameplay a lot better and like make it nicer for players out there so they

know AI and gaming is not all about uh you know throwing people out of jobs or whatever people like to think these days. There's also trying to help people be able to play uh more fair matches and things like that. So
cool.
No, thank you very much for the conversation. It was nice to talk about these topics because sometimes they are not easy and uh I can say that finally people is also paying more attention to that. It was not the case on the pass and also thanks as well for for the diamond interest.
Awesome. Well, thanks again for for stopping in and uh and thanks to everyone out there listening. Uh we'll catch you guys on the next one. If you enjoyed today's episode, whether on YouTube or your favorite podcast app, make sure to like, subscribe, comment, or give a fivestar review. And if you want to reach out or provide feedback, shoot us a note at [email protected] or find us on Twitter and LinkedIn. Plus, if you want to learn more about what Navic has to offer, make sure to check out our website, www.navic.co. There you can sign up for the number one

games industry newsletter, Navic Digest, or contact us to learn about our wide-ranging consulting and advisory services. Again, that is www.navic.co. Thanks for listening and we'll catch you in the next episode.