FACEIT just escalated the arms race against cheaters. The competitive Counter-Strike 2 platform is rolling out a machine-learning system called “Human Input Detection” that doesn’t look for known cheat software. Instead, it watches how you play and decides whether a human being is actually behind the keyboard.

The feature launches August 5, 2026, timed with the start of Season 9. It represents a fundamental shift in how FACEIT approaches anti-cheat: rather than scanning for software signatures, the system analyzes in-game inputs to determine if they’re authentically human.

Why traditional anti-cheat is losing the battle

Conventional anti-cheat systems work like antivirus software, maintaining a database of known threats and scanning for matches. AI-powered cheats exploit exactly that weakness. These tools don’t inject code into the game or modify memory in ways that leave obvious fingerprints. They read screen data through external hardware (often via DMA cards) and translate it into seemingly legitimate mouse movements. To a signature-based scanner, everything looks clean.

The numbers tell the story of how bad this has gotten. As of May 2026, AI and DMA cheats accounted for 40% of all bans on FACEIT’s platform. During Season 7, AI cheat detections spiked 272%, resulting in 372 bans.

FACEIT’s existing kernel-level anti-cheat, which operates at the deepest layer of a computer’s operating system, improved its coverage of AI-supported cheats through mid-2026. But kernel-level access alone wasn’t enough.

How Human Input Detection actually works

The new system is trained on millions of matches worth of input data. It studies how humans actually move a mouse, click, and adjust aim, then builds a model of what “normal” looks like. When inputs deviate from human patterns in specific, telltale ways, the system flags them.

FACEIT has been working with Google engineers on behavioral modeling since early 2026, which helped build the foundation for this detection layer.

One critical design choice: the system requires multiple confirming signals before any ban is executed. A single suspicious input sequence won’t get you kicked. FACEIT has also built in specific accommodations for accessibility devices. Players who use adaptive controllers, alternative input methods, or assistive technology shouldn’t trigger false positives.