Dataset and AI in Detecting Player Cheats in FPS Games

The evolution of cheating in first-person shooter (FPS) games has gone from simple memory editing to sophisticated uses of AI, which are major threats to the fairness of games and the player experience. As competitive gaming and esports grow rapidly, the social and economic impacts of cheating continue to increase, and studies show potential player churn rates of 78%. Cheating was typically addressed using traditional anti-cheat systems that used a signature-based detection method. However, the use of computer vision and hardware assistance by modern cheats allows them to bypass traditional detection methods. This review examines methods of artificial intelligence and machine learning to detect cheaters in FPS games as a result of a thorough review of literature articles that have explored various methods of detecting cheaters, which include XGBoost, random forest, deep learning architectures like CNN and LSTM, and transformer-based models. The findings show that space-temporal analysis hybridized models have detection accuracies of over 99%. Moreover, even the unsupervised forms of cheat detection are potential in detecting some new forms of cheats. The boundary between human and AI cheats will always be a challenge to the anti-cheat systems because of the possibility of humans being labeled as elite.