Yes, a lot of anti-cheats now look at how you aim and move in a match, on top of scanning your PC for cheat programs. It's called behaviour detection or input detection, and FACEIT, Riot, Activision and PUBG have all said publicly that they use some form of it. This post covers what it looks at, which games say they use it and what it means if your aim goes through a MAKCU or FERRUM.
What behaviour detection looks at
A normal anti-cheat looks for cheat software, drivers or devices on the PC you play on. Behaviour detection looks at the match itself, at how your crosshair moves and how you react. That means things like how fast your aim gets onto a target, how it speeds up and slows down on the way, how long you take to react when someone shows up, whether you overshoot a bit and pull back like people do, and how consistent all of that stays from one fight to the next.
People are messy when they aim. Reaction time changes when you're tired, flicks miss, and a bad day looks different from a good one. These systems are trained on huge amounts of normal gameplay, so they learn what that mess looks like and flag play that sits outside it, usually over lots of fights and matches instead of one clip. None of the companies say exactly which signals they use, since that would only help cheat makers, so anything more detailed than this is a guess.
Games that say they use it
FACEIT
FACEIT added Human Input Detection to its anti-cheat with Season 9, which went live on August 5, 2026. FACEIT describes it as a machine learning layer that catches cheaters by what they do in the game, trained on millions of matches. Its FAQ says a flag gets weighed together with the rest of their anti-cheat before a ban, and that accessibility devices count as normal human input.
In its September integrity update, FACEIT said Human Input Detection had looked at 6.3 million hours of play and helped their team ban 70% more cheaters than in the same stretch of Season 8. That's a count of bans and says nothing about how often it gets it right or wrong. None of our firmware works on FACEIT anyway.
VALORANT
Riot lists "Behavior" as one of the reasons it bans people in VALORANT. In a February 2025 post it calls that a machine learning suspension, also known as server-side anti-cheat, mostly given to ragehackers. Server-side means it runs on Riot's servers from match data, so it doesn't need anything on your PC to judge how you play. Riot's Vanguard On-Demand post says this system mostly removes ragehacker accounts instantly, so blatant play can get caught fast.
Call of Duty
Before Black Ops 7 came out, the RICOCHET team wrote that its aimbot models are trained to tell whether a player's aim came from a person or from an aimbot. Other models look at how a player moves, fights and changes over time, and flag things like aim that snaps across the map faster than a person could follow.
PUBG
In a November 2025 dev letter, PUBG's team said it runs an AI system that finds and bans abnormal behaviour and goes further than looking for cheat software, and that it feeds into their video reviews. In April 2026 they added extra checks to it to cut down on false bans, and said they're rolling out new detection for mouse scripts and for devices that send abnormal input.
What it means if you use a MAKCU or FERRUM
With an aim device, your software runs on your second computer and your main PC only sees a normal mouse, so a scan of the main PC has no cheat program on it to find. That's what the device changes, but the game still gets every movement that comes out of it, and behaviour detection judges those movements. Inhuman aim looks just as inhuman coming from a MAKCU or FERRUM as it does from a program on your PC, so a device doesn't make it safe.
Behaviour detection is also usually one signal out of several. FACEIT weighs it with its other checks, Riot lists it next to other ban reasons, and PUBG runs its AI results through extra checks and video review. Reports, stats over many matches and hardware checks all add to it, so it's rarely one odd fight that decides things. Blatant play is the exception, since Riot says its system mostly bans that on the spot.
What about humanization
Some cheat software tries to make automated aim look human. The general idea is to copy how real people move, for example by learning from recordings of real players' mouse movement, with the pauses, speed changes and small corrections in it, or from the way you move yourself.
That can make detection harder, but nobody outside the anti-cheat teams knows by how much, and those teams train their models on real players too. The moment assistance kicks in, the change from how you were moving a second earlier can still look unnatural. Strong assistance is easier to spot than light assistance, because the further the aim gets from what you could do yourself, the more it stands out over lots of fights. Anti-cheats also put behaviour together with reports, stats and everything else they check, so no setup is undetectable.
What we tell you
Any build can get detected at any time, and how you play gets checked on every setup, so we'll never promise you won't get banned and we ask you to play responsibly. Is DMA detectable? covers the other ways people get caught, and the status page shows how every device is doing right now.
You don't need an account, this tells us what to write more of.