A review of new science, in plain language

Nature  ·  30 September 2026  ·  From issue 46

Scalable decision-making for games of imperfect information

An AI beat the top human Stratego champion, a first for the game.

Samuel Sokota, Eugene Vinitsky et al.

An AI program called Ataraxos beat the best human Stratego player ever, in a board game full of hidden information. It learned by playing itself over and over, then planned each move while keeping its own secrets safe. The same method also built strong AI players for other hidden-information games, including a cooperative card game and a team card game, using far less computing power than past attempts.

Integrity screen: passed (3 checks) Checked 6 October 2026. Read the source

The plain-language reading of this paper is in the full issue, for subscribers.

Read the full issue

Five vetted papers in plain language, free, every two days

Five papers free every issue, about 15 emails a month. No spam. Leave in one click.

Also in issue 46 15 papers
More from Nature