MIT's Ataraxos Masters Stratego, Beats Top Human Player

On September 30, 2026, MIT News reported that a new AI system built by MIT researchers had beaten top-ranked human players at the strategy board game Stratego, and that it was more efficient than other models. The researchers suggested such systems could help decision-makers in military maneuvers or business negotiations. That report did not give the system a name.

The following day, The Decoder identified it as Ataraxos, describing it as the work of researchers from Carnegie Mellon, NYU, Stanford and MIT, built for less than $8,000. According to that report, Ataraxos decisively beat the best Stratego player of all time. Stratego is a hard test for AI because both sides set up their pieces face down, and the report noted that Google DeepMind fell short in 2023 despite a multimillion-dollar budget. The two accounts differ on attribution: MIT News credits MIT researchers, while The Decoder lists four institutions.

On October 2, 2026, Ars Technica reported what made the win possible: a second neural network whose job is to guess the identity of the opponent's hidden pieces. That is where the public record currently stops. The system has beaten the strongest human Stratego player, and the key technical addition is described as a second network dedicated to inferring hidden piece identities; further details have not been disclosed.