Riot Games uses reinforcement learning and imitation learning to train AI-controlled bot opponents that players can compete against in League of Legends and 2XKO, aiming to create more realistic and challenging practice opponents than traditional rule-based bots.

Riot's ML Bots team, part of the AI Foundations group, develops game-ready AI agents using methods including reinforcement learning, imitation learning (behavior cloning), and policy gradient methods. For its fighting game 2XKO, Riot's Head of Technology Research publicly described a structured process where a research team first de-risks learning agents before handing them to a central tech team to productionize, with live game teams monitoring bot behavior post-launch. In League of Legends, Riot released a major overhaul of the bot framework in patch V14.6 in 2024, enabling more nuanced AI decision-making. The team also uses player telemetry data and gameplay data to guide bot behavior design.