VACnet is Valve's deep learning system that watches how players move and aim in Counter-Strike matches and automatically flags those whose behavior looks statistically impossible for a human. Valve first revealed the system at a game developers conference in 2018. It processes roughly 150,000 matches per day and routes flagged cases to human reviewers rather than issuing bans directly.

VACnet uses a deep learning model trained on labeled data from Valve's "Overwatch" system, where experienced human players reviewed match replays and voted on whether someone was cheating. The model captures 140 individual data points — called "atoms" — from an 8-round window of gameplay, focusing tightly on changes in crosshair position immediately before and after each shot. When deployed in a 2-versus-2 competitive mode, VACnet achieved a conviction rate of 99% on cases it flagged. Valve engineer John McDonald stated at the time that certain cheat types had been reduced by a factor of one hundred. The system runs on approximately 1,700 CPUs continuously. Valve filed US Patent 20190291008 covering the deep learning approach to cheat detection.