Cheating practices are becoming more common as a result of internet gaming’s explosive expansion, endangering the fairness of competitive settings. To ensure fair play, improve player experience, and spot such behaviors early, real-time anomaly detection is essential. The main goal of this study is to identify unusual player behavior by analyzing gameplay data with a bespoke keylogger and machine learning algorithms. Even with all of the advancements in anti-cheat systems, it is still difficult to achieve high detection accuracy without producing too many false positives. It is challenging to apply many of the current approaches across various gaming platforms due to their issues with scalability, adaptability, and real-time performance. In order to permanently adapt to the players’ ways and alert them whenever something out of the ordinary happens, we propose a new approach based on the analysis of keystroke and mouse movement patterns using the Isolation Forest model. We are different in a way that we focus on real-time, low-latency anomaly detection without ignoring the research challenges that have to do with feature engineering and minimizing false positives. The results of experiments have shown that our model is much more precise and recall-based than other models, thus ensuring more accurate results. There is an improvement in the real-time detection performance of the strategy and the level of false positives has also significantly reduced compared to the standard procedures. This work offers an architecture for detecting anomalies in computer games that can easily be scaled up and is effective in other interactive applications.