Google DeepMind is partnering with the creators of the long-running online game EVE Online to train and research advanced artificial intelligence systems in highly complex virtual environments. The initiative highlights how AI companies are increasingly turning to sophisticated multiplayer games as testing grounds for next-generation intelligent agents.
As part of the collaboration, Google DeepMind has reportedly taken a minority stake in CCP Games, the studio behind Eve Online, which has recently rebranded as Fenris Creations. The partnership will allow DeepMind researchers to study player-driven systems, economic interactions, strategic behaviour, and large-scale coordination within isolated versions of the game environment.
Eve Online is considered one of the most complex virtual worlds ever created, featuring player-run economies, political alliances, wars, logistics networks, and long-term strategic planning. Researchers believe these characteristics make it an ideal environment for training AI systems that require skills beyond simple gameplay, including continual learning, negotiation, adaptation, and resource management.
The project builds on Google DeepMind’s long history of using games to advance AI research. Previous breakthroughs included systems such as AlphaGo, which defeated world champions in the board game Go, and AlphaStar, which mastered the real-time strategy game StarCraft II. More recent initiatives like Genie and AlphaEvolve demonstrate DeepMind’s growing focus on world models and autonomous problem-solving systems.
According to the companies, the research will initially take place on offline, controlled versions of Eve Online rather than the live game servers. Beyond research, the collaboration may also explore new AI-powered gameplay experiences in the future. Industry analysts view the move as part of a broader trend where video games are becoming important laboratories for developing artificial general intelligence systems capable of operating in dynamic, unpredictable environments.
