How Anybrain is Using AI to Fight Video Game Hackers

"Our goal was not to create an anti-cheat; it was to create a product that works."

—Anh-Vu, the Chief Growth Officer (CGO), Anybrain.

Anybrain is a Portugal-based company that is currently popular for its AI anti-cheat solution for online games, co-founded by André Pimenta and Serafim Pinto.

André Pimenta Ribeiro (PhD in Computer Science/Artificial Intelligence, from the Universidade do Minho) was the visionary of Anybrain. He is an entrepreneur and software engineer dedicated to the gaming industry, with over 9 years of experience in software engineering and development. Ultimately, he is a researcher focused on non-intrusive analytical methods for behavioural patterns, which most of the company’s products are based on.

Before their recent success and breakthroughs in 2022, Anybrain –Co-founded by André Pimenta and Serafim Pinto– has been working hard in the background since 2015.  You’re probably wondering what they were doing before they became this popular. Apparently, they were just focused on building a product that works, but it led them here. Their story is the typical case of following the flow and letting it take them somewhere great. It reminds me of Bruce Lee's infamous quote, "Be water, my friend."

Interestingly, their initial mission wasn't to create an AI anti-cheat. Anybrain is an innovative game solution and DevOps company that uses analytics to drive fairness in game systems, no matter what platform or kind of game. It has the potential to be the solution to the future of gaming security. Theirs is a story of building a company from passion, but it also has industry cum market awareness and a team of similar minds.

In my interview with Anh-Vu Nguyen (CGO) and Serafim Pinto (Co-Founder and Chief Technology Officer) of Anybrain, they spoke about their interests in developing a product, the gaming industry, market dynamics, and the future of the company. The interview covers details of their operations and technology in the video game industry, which is an open sea of multiple careers, means of livelihood, and an outpouring of emotional sentiments.

Lessons for Tech Companies:

Here are some lessons to look forward to from the interview, which will help tech companies.

• The success of a tech company, especially if using generative AI, relies greatly on its ability to be innovative and remain trendy while meeting consumers' needs. There is no escaping this for companies leveraging generative AI.

  
  

• We have seen tech products remain dormant and die a natural death because they want to stick to a product or a pattern of doing things; for instance, Blackberry, Nokia, and Kodak.

  
  

• In a world where people are more concerned about security, less-invasive solutions will always win people over.

  
  

• Having a team passionate about performance is an advantage.

  
  

• Consistent, not just initial, market research involving awareness, consumer demographics, and behaviour helps relevance in an ever-adapting world.

The Interview

Q: Hello, it’s nice to e-meet you. Can you tell me about yourselves and Anybrain?

Anh-Vu: Anybrain is a game solution technology and platform that uses AI analytics and deep learning to offer a fair and enjoyable gaming experience to online users.

Serafim: André was doing his PhD study and working on developing a generative AI model that identifies mental stress based on how people use their PC and mobile devices. Later, I joined him, and we worked together to develop a solution to address the issue.

The solution technology from that time became the background for what would be known as Anybrain's main technology today: the analysis of player behaviours to draw conclusions and offer actionable insights to game studios and developers.

Hence, although we began to offer in-game solutions in 2015, we have been working in the background even before then, building the technology that forms the basis of our main product.

Q: Does everyone have a background in computing and programming?

Serafim: "Yes, everyone on the team does. André has a PhD in AI studies, and I met him around that time as a computer science student myself. Everyone has a background in computer science.

Anh-Vu: Everyone has a background in computing, but more importantly, everyone has an interest in gaming. So, everyone here is a gamer. I am the marketer and businessperson on the team, but I still have a MSc. in computer science. So, I understand a thing or two about AI and computing, but not as much as the other guys. So, yeah, we all do.

On top of that, we are all gamers.

Q: You have existed since 2015. I read that you began working with eSports some years, but that is the farthest I've seen. What technology solutions have you been offering since then? Have you always focused on games?

Serafim: Here’s some more background on Anybrain.

During André’s Ph.D. research, he researched and developed AI algorithms and technology to study mental fatigue patterns. After that, he and the budding team conducted research on mental fatigue among eSports players. We proceeded to develop software to detect mental fatigue and stress levels in eSports games.

We were using this to help teams optimise their players and know how to manage them for their games. We called this solution Performetric. It is a non-invasive programme that runs on PCs to analyse mental fatigue and stress based on how players click their mice and keyboards.

So, in our early days, that was in 2018, we were serving eSports teams to help them monitor their players' stress and anxiety levels so they could know what to do. While doing that, we were also looking for other ways to expand our reach.

Eventually, as we analysed player behaviours and their use of their hardware, we started the solution that is now known as Anybrain’s AI Anti-Cheat. We spent the earlier years testing the products and then pitching prospects to show them how our solution works and why we are better than other existing ones in the market. We were trying to gain recognition for what we were doing. Seeking that recognition took a while, but it has brought us here eight years later.

Anh-Vu: So, yes, we have existed as a team for a long time. But I joined them more recently. André and Serafim began working on the technology for our line of products many years ago.

Q: In these 8 years, what are the milestones the company has achieved that you believe have led you to this point?

Anh-Vu: First, developing a solution that works and proves efficient in 99% of cases is the first milestone. That took André and Serafim almost 8 years and it will keep on evolving.

Then, after COVID-19, we could attend game shows again. So we did most of them in 2022, and it was a game changer for us. We got to talk to potential customers, in person this time, and start building our reputation that way. We got our first 5 customers, a bit of esports and small studios. This helped a lot in building something that would be up to the expectations, and even more so, avoid disappointment from the community and industry. And on this topic, this happened a lot in the past.

In 2023, the next milestone was to show that, what we could do on a smaller scale, we could also do at AAA scale, which we achieved early 2023 with a successful test on a big MMOs.

Now, we have a few live customers, on PC and mobile, and we are preparing to go live with multiple big titles, on consoles as well this time.

Q: In my conversation with Ricardo (the digital content manager), he says you are "really working on a solution that is ready to be implemented in all types of games and platforms." Does this mean you are still working on the technology, or has the solution been completed as ready-to-use or adopted by your clients with different use cases?

Serafim: Our solution is ready and available. However, we work with different game studios whose systems and platforms often vary. For instance, FPS games are similar in play, but the platforms are different. So, we still have to learn the systems of our clients and train our ML model to analyse the dynamics before we can fully integrate our solution.

So, you can say our solution is up and running, but it evolves and adapts. We continue to adapt to our clients' needs by creating new models for their respective systems, even for existing clients, every time they launch new products since the games are different.

Anh-Vu: Also, in game development, there are different levels of security. The primary level is where game developers protect their systems for the safety of users, protecting their identity and information from any loopholes hackers can use to extort players.

There is secondary security, which is about creating a fair and level playing field for users, including the prevention of cheating and manipulation. So when the hackers breach the client, you need a secondary layer of security like Anybrain to detect fraudsters in real-time.

Then there are even extra layers of security like telemetry analysis and toxicity moderation.

Q: Some people think they might still be able to cheat Anybrain's AI by not playing all their cards. E.g., since they know what parameters Anybrain looks out for, they might decide to play along, yet use the advantage of their cheats to kill other players and accomplish missions. Is it actually possible to cheat Anybrain, and if so, how will you combat this?

Serafim: This is somehow addressed in the previous answer. We are constantly working on and feeding our models. We are not in a rush to catch cheaters. In fact, our goal for Anybrain is not just to create an anti-cheat. It is to create a level playing field for online games and the gaming experience in terms of safety and fairness.

Anh-Vu: We are not aiming to catch all hackers and cheaters. That is because we set our threshold really high for the sake of integrity. So, we may not catch everyone cheating, but the ones we flag, we are sure they are cheating.

In fact, we have had some clients tell us to reduce our threshold in certain instances so they can see more suspects. We don't ban; we only detect foul play, leaving the studios to take action.

Q: While researching and following up on discussions concerning your impressive AI, some recurring comments were about not being able to use people's accounts. E.g., I am visiting a friend and I cannot use his account because I could be a better gamer than he is. If your system notices this sudden improvement, it will flag this account as a cheater. Is this concern a valid one?

Serafim: So, we understand that these things happen. However, we are only responsible for the detection of cheating systems. We only detect foul gameplay and suspected cheating, and we report such profiles and suspected cheats to the studios through the dashboard. The studio has the final say on who gets banned or not. We only flag accounts based on criteria identified by our programme, and the account is reported to the studios.

Sometimes, we report some accounts, and a game studio may do nothing about it. Sometimes we are asked for recommendations, and we can give our honest opinions.

Something to know is that we are not too quick to flag a player; we consider a player's growth and adaptability to the system over time.

So, a random play with good reflexes, aims, skills, and other cognitive capacities may not be flagged. It is also why we seem to be careful in catching cheaters, but once we flag them, we are certain that they are cheating.

Again, we flag certain accounts or profiles for cheating, but we leave the companies to decide whether to ban them or not.

Interesting! That sounds like blowing a whistle but not being responsible for the fall.

Anh-Vu: We understand that these things happen and that’s why we build player profiles to see if how they play has been consistent with their growth and familiarity with the game.

In fact, for multiplayer games, we are equally focused on smurfing (creating new accounts) for the purpose of cheating.

So, common actions taken by most companies from our report are that when you (a player) use rapid fire, we will always signal a warning, and such players will be on a watchlist. But using an aimbot earns a direct ban.

Q: On your Twitter (now X) page, I see an image that says you are on 30 games. But that was in January. If so, can you mention some games using Anybrain? How many cheaters have you been able to catch within a specific time frame—say, in the last year or six months? If you are not able to disclose the games at this time, can you talk about the game categories? E.g., action such as FPS, adventures, or board games?

Ahn-Vu: Yeah. On mobile games, we are present in more than 30 games. We cannot announce them because that is not our aim. Our aim is to have a solution that works, and we do a better job when we are anonymous or people don't know the kinds of games we work with because cheaters do not know what platform we are on and speculate how to hack.

Our partners may decide to announce us, and that is totally at their discretion. For us, anonymity is important, but the aim is security.

So, apart from FPS, we are currently integrated into racing games and card games ( solitaire-like), and we have started to talk with chess platforms. Many cheats exist on chess platforms!

Sometimes, we just have a single-player platform client approach us. But in the coming months, we will be releasing more games for PCs, mobiles, and even consoles.

Q: Just a follow-up question: since your customers are anonymous, it must be difficult to show prospective clients how well your products work. So, how do you convince new customers?

Anh-Vu: Demonstrations are the primary way to convince customers. We demonstrate to them that our solution works, and when they see it, they can decide if they want to work with us. That is why attending gaming events is a big deal for us.

Q: Wow! That's huge. Did you leverage existing technologies or solutions on the market while building your products? e.g., what they were doing, how were they doing it better than their competitors, and how would you do it better?

Serafim: Based on how we started, there were solutions on the market, but they didn't do what we do. There were no non-invasive tech solutions. Existing ones used high-resolution cameras that analysed eyes; some require users to have the programme on their computer, which scans all their files for hacks. Some people even use keystroke dynamics to detect the genuineness of gameplay.

Ours is unique because we were looking forward to a passive way of detecting cheats. We are not scanning client files, which means we don't expose their systems to bugs.

We only collect the data needed to interpret player interactions in-game, and it is immediately encrypted.

We are also working on a beta programme that can determine age based on how people use their devices. It is still in the R&D stage. This technology is to give the right information to the studios/developers. By knowing the actual age of gamers, they can decide whether to act or not on certain situations involving children, as well as adapt content depending on this.

Anh-Vu: I mean, when we were younger, we all lied about our age and consented that we were above a certain age to enjoy some games. So, we are working on that, too.

Bearing in mind that the mission is not to create an anti-cheat but to provide security and fairness, Anybrain can expand its product line using a similar model of biometric profile or fingerprinting through user analytics.

Q: That’s impressive. I read somewhere that you have a patent for your technology. I'd like to know which of the solutions that you offer is patented.

Anh-Vu: Yes, we have two patents. We have one covering Europe and the other covering other parts of the world. It's a UPN for the method of collecting and building data. Those are unique.

Q: I see. Does your deep learning model use supervised data (manual entry of information for existing images or files) or both supervised and unsupervised (no manual annotations are done, but leveraging an existing information bank for a file or image) data, given how much progress you have made?

Serafim: We use supervised and unsupervised data to train our models; however, we don't use images or files. Sometimes we get data directly from the studio, and sometimes we create the hacks ourselves.

Q: What are the biggest challenges you currently face, and how are you addressing them?

Anh-Vu: Our major challenge in the past was being unknown. It is very difficult to address gaming security when no one knows you. However, we have moved beyond building a reputation. Our reputation is proven, but we have more to worry about.

Building a company on anti-cheat is the problem. It is difficult because cheaters are always evolving, and you have to be ahead of them. So, we are constantly faced with the question, "For how long will we be relevant?"

Our relevance depends on our ability to anticipate what hackers and cheaters will do. Besides, it is only a matter of time before similar technologies come up with, perhaps, even more intriguing and innovative solutions.

Moreover, due to anonymity, getting recognised early is a challenge, and sometimes some games where we are integrated delay launch; we only officially begin to work for them when they have players. Sometimes it takes months for companies to roll out their products.

So, what we do in this case is evangelise our products and their credibility to companies.

Serafim: Hardware hacks are also more challenging to tackle because you don't know for sure what tweaks gamers have done to their gears. However, I project that hardware manufacturers will also be incorporating anti-cheat solutions into their gears. Consoles will be developed with more security in mind.

Q: Finally, what does the near future look like for Anybrain? Any exciting news to tease?

Anh-Vu: You can anticipate us coming out as integrations with some new games and big studios, especially AAA games. We will be in many of them.

We are working on an age verification feature where we can detect whether children are playing games beyond their age and mixed with a population that shouldn't be in such a "close" digital proximity. We are working on games for children, and we will be analysing their play styles and patterns for some other games.

We are hoping to track cheaters for all kinds of hacks, just as we do for aimbot users.

We also have other R&D projects in the early stage. They include:

• Bots with more human behaviours, the "reverse-anybrain" to populate lobbies of dying games.
• The use of our 'mechanical skills' approach to better improve  skill-based matchmaking (SBMM).

Conclusion

André Pimenta and team’s story is one of creating a multi-million dollar company with a product goal and passion for performance software development, both of which drive necessary products according to the awareness of the market.

Anybrain, as a company, has done well for itself. However, it still has to roll out more unique products to maintain its relevance in the industry. That is because the AI anti-cheat niche in the video game industry faces a double challenge: hackers who can make a solution irrelevant, and the out-performance from competitors. It must work faster to extend its line of products to serve other areas of gaming to stay relevant. I trust they will.

Yet, it has an advantage: a unique technology that has helped its products (ones made publicly available and ones underway) and a team of enthusiastic and performance-driven staff who share the same gaming passion. I believe they will do well and thrive in the next ten years if they quickly shift their weight from a single product to more products that more games can take advantage of.

I am blown away by the company's creativity and technology. My favourite is how analytics are run through hardware and on-screen actions in a non-intrusive way. I am an ambassador for products that work. As long as their technology keeps impressing, I'll remain a fan. We can only hope the company keeps delivering on its promises.

The future is shaky but promising. Isn’t that the risk for businesses and enterprises anyway?

**Special thanks to Ricardo Silva, Serafim Pinto, and Anh-Vu Nguyen at Anybrain for making this interview possible.