AI made cheating easy for everyone, so anti-cheat learned to watch how you move

Ask almost any player and they will tell you their favorite game is full of cheaters. Most of the time they cannot actually prove it, and that uncertainty is the whole problem. Andre Pimenta, CEO and co-founder of Anybrain, has spent roughly eleven years building an answer that starts from an unusual place: it never looks at your game at all. It watches how you move.

Anti-cheat that never looks at your game

Most anti-cheat scans the game's memory, hunting for something tampering with the code. Anybrain does the opposite. It reads a player's inputs, the mouse, the keyboard, the controller, the exact rhythm of clicks and movement, and builds a behavioral signature from them. The approach grew out of Pimenta's PhD, which detected mental fatigue from the way people used a computer. The insight carried straight into cheating: humans are not random, so the signature holds even when the on-screen gameplay looks perfectly normal.

That is why an obvious cheat gives itself away. When a mouse jumps from one point to another without the movement that should connect them, the replay looks fine, but the inputs are missing the steps a human hand would have made. Anybrain now monitors millions of players this way, looking for patterns that are invisible to the eye and even hard to spot in the raw data.

AI erased the barrier to cheating

The uncomfortable shift Pimenta describes is that cheating used to take a little effort, and now it takes almost none. A player once had to hunt through a forum or a GitHub repository to find a script. Today an LLM will write that script on request, and cheap computer vision reads the screen to figure out where the enemy is and where the mouse should go.

It is not a PC-only problem anymore either. For years the assumption was that consoles could not be cheated. Pimenta points to a wave of new devices that inject or modify input from outside the game: they run computer vision on the screen and feed commands through hardware pretending to be an ordinary controller, so the cheater never touches the game's code. The tools got cheaper, the knowledge got easier, and the volume is going up.

The real damage is doubt, not the loss

For the operators running live games, Pimenta's argument is that detection alone is not the fix. The deeper damage is that most players genuinely cannot tell whether they were beaten by a cheater or simply by someone better, and that doubt corrodes trust in the game itself. Worse, cheaters actively manufacture noise, insisting they are innocent and that the game is broken, to make the whole community stop believing the system works.

His prescription is transparency. The studios handling it well now publish their numbers, how many bans, what kinds, on a regular cadence. Anybrain worked with Crytek on Hunt: Showdown, and Crytek publicly shared the console detection figures from the first month. When players can see the system working, they stop assuming everyone across the lobby is a cheater. The catch is that anything shared with the community is also read by the cheaters, so how much to reveal is a genuine balance.

Checking a player's age without ever seeing their ID

The same behavioral engine reaches past cheating into one of the industry's hardest current problems: age verification. Almost everyone does it with an ID upload. Anybrain does it from behavior. The way a person handles a mouse, keyboard, controller, or phone tends to fall into an age band, so on a server meant for kids, inputs that look like an adult get flagged, quietly, to the publisher.

Crucially, the system does not announce that a player is a specific age or connect them to a real-world identity. It only says this account is worth a second check. That reflects a principle Pimenta holds deliberately: a player's game identity and their real identity should stay separate. It is safety that respects the line rather than demanding an ID at the door.

Why humans still make the call

For anyone who works trust and safety or community, the reassuring part is that a cheat cannot hand you skill. It can give you recoil control or auto-aim, but not game sense, which still takes hours and real decision-making, so plenty of cheaters remain bad players with a crutch. And Anybrain is adamant that detection cannot be a black box. It builds explainability tools so a community manager reviewing a ban appeal can look at the actual evidence, the inputs, the drivers, the history, and understand the call.

In fact, Anybrain does not ban anyone. It hands the studio the proof and a human makes the decision. In an era where AI is arming both sides, that human-in-the-loop, evidence-first posture is the point, not an afterthought.

Watch the full conversation with Andre Pimenta on YouTube.

Want the operator's view on trust, safety, and live games every week? The Operator's Briefing goes out every Tuesday: https://playerdriven.io/level-up . And the conversation continues in Discord: https://discord.gg/GC5PkbKFH .