--- title: With most information hidden, the game Stratego had... description: Researchers from Carnegie Mellon, MIT, NYU, and Stanford built an AI called Ataraxos that defeated Pim Niemeijer, considered the best Stratego player ever, 15... canonical: https://daily.dev/posts/with-most-information-hidden-the-game-stratego-had-stumped-ai-until-now-gs01dniuf twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: With most information hidden, the game Stratego had stumped AI—until now | daily.dev og:description: Researchers from Carnegie Mellon, MIT, NYU, and Stanford built an AI called Ataraxos that defeated Pim Niemeijer, considered the best Stratego player ever, 15... og:url: https://daily.dev/posts/with-most-information-hidden-the-game-stratego-had-stumped-ai-until-now-gs01dniuf og:image: https://api.daily.dev/og/posts/Gs01DNiuF.png og:image:alt: With most information hidden, the game Stratego had stumped AI—until now og:image:width: 1200 og:image:height: 630 og:locale: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # With most information hidden, the game Stratego had stumped AI—until now **[Ars Technica](https://daily.dev/sources/arstechnica)** · 2 min read · 0 upvotes · 0 comments ## Summary Researchers from Carnegie Mellon, MIT, NYU, and Stanford built an AI called Ataraxos that defeated Pim Niemeijer, considered the best Stratego player ever, 15 games to one with four draws. Unlike chess or Go, Stratego hides piece identities and can last thousands of moves, with a decillion possible setups, which stumped prior systems including DeepMind's DeepNash. The breakthrough came from adding a second neural network that infers the likely identity of hidden opponent pieces, and the model was trained using just 16 GPUs and a few thousand dollars, far less than DeepMind's effort. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### Has an AI ever beaten a top human player at Stratego? Yes, an AI called Ataraxos, built by researchers from Carnegie Mellon, MIT, NYU, and Stanford, defeated Pim Niemeijer, regarded as the best Stratego player of all time, by a score of 15 games to one with four draws. It was trained using only 16 GPUs and a few thousand dollars, a far smaller budget than DeepMind used for its earlier DeepNash system. _Developers tracking milestones in game-playing AI can follow breakthroughs like this on daily.dev._ ### Why is Stratego harder for AI than chess, Go, or poker? Stratego involves a massive amount of hidden information that unfolds over very long games. Each side has 40 pieces whose identities are secret until battle, producing more than a decillion possible setups, compared to just 1,326 possible hands in Texas Hold'em. Games can last roughly 2,000 moves versus about 40 in chess, and effective bluffing adds further strategic complexity. _Anyone comparing AI approaches across games can dig into these tradeoffs on daily.dev._ ### What technical approach let Ataraxos solve Stratego's hidden information problem? Ataraxos adds a second neural network specifically designed to guess the identity of an opponent's hidden pieces, working alongside the primary decision-making network. This addition was the key innovation that let it handle Stratego's massive hidden-information space, something earlier systems like DeepMind's 2022 DeepNash struggled to fully crack. _Engineers curious about imperfect-information AI design can explore more coverage on daily.dev._ ## Similar posts on daily.dev - [Games people — and machines — play: Untangling strategic reasoning to advance AI](https://daily.dev/posts/games-people-and-machines-play-untangling-strategic-reasoning-to-advance-ai-hhv1dr2uv) · MIT News · 0 upvotes · 0 comments - [AI is rewiring how the world’s best Go players think](https://daily.dev/posts/ai-is-rewiring-how-the-world-s-best-go-players-think-g6jbvmfxc) · MIT Technology Review · 0 upvotes · 0 comments - [Which AI Lies Best?](https://daily.dev/posts/which-ai-lies-best--taqhbbdck) · Hacker News · 2 upvotes · 0 comments --- Tags: [#ai](https://daily.dev/tags/ai), [#neural-networks](https://daily.dev/tags/neural-networks), [#reinforcement-learning](https://daily.dev/tags/reinforcement-learning) [View this post on daily.dev](https://daily.dev/posts/with-most-information-hidden-the-game-stratego-had-stumped-ai-until-now-gs01dniuf) ```json {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]} {"@context":"https://schema.org","@type":"TechArticle","headline":"With most information hidden, the game Stratego had stumped AI—until now","url":"https://daily.dev/posts/with-most-information-hidden-the-game-stratego-had-stumped-ai-until-now-gs01dniuf","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/with-most-information-hidden-the-game-stratego-had-stumped-ai-until-now-gs01dniuf"},"datePublished":"2026-10-01T16:49:51.879Z","dateModified":"2026-10-01T16:51:38.291Z","description":"Researchers from Carnegie Mellon, MIT, NYU, and Stanford built an AI called Ataraxos that defeated Pim Niemeijer, considered the best Stratego player ever, 15...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/cc9d1dcfdd2578cc03c23fedadfdbe83?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/cc9d1dcfdd2578cc03c23fedadfdbe83?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"Ars Technica","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"Ars Technica","logo":"https://media.daily.dev/image/upload/t_logo,f_auto/v1/logos/80883e0e48a34b5ebcf93777016cb3fe","url":"https://daily.dev/sources/arstechnica"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/with-most-information-hidden-the-game-stratego-had-stumped-ai-until-now-gs01dniuf","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":0},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"ai,neural-networks,reinforcement-learning","timeRequired":"PT2M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Ars Technica","item":"https://daily.dev/sources/arstechnica"},{"@type":"ListItem","position":3,"name":"With most information hidden, the game Stratego had stumped AI—until now"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/with-most-information-hidden-the-game-stratego-had-stumped-ai-until-now-gs01dniuf#faq","mainEntity":[{"@type":"Question","name":"Has an AI ever beaten a top human player at Stratego?","acceptedAnswer":{"@type":"Answer","text":"Yes, an AI called Ataraxos, built by researchers from Carnegie Mellon, MIT, NYU, and Stanford, defeated Pim Niemeijer, regarded as the best Stratego player of all time, by a score of 15 games to one with four draws. It was trained using only 16 GPUs and a few thousand dollars, a far smaller budget than DeepMind used for its earlier DeepNash system. Developers tracking milestones in game-playing AI can follow breakthroughs like this on daily.dev."}},{"@type":"Question","name":"Why is Stratego harder for AI than chess, Go, or poker?","acceptedAnswer":{"@type":"Answer","text":"Stratego involves a massive amount of hidden information that unfolds over very long games. Each side has 40 pieces whose identities are secret until battle, producing more than a decillion possible setups, compared to just 1,326 possible hands in Texas Hold'em. Games can last roughly 2,000 moves versus about 40 in chess, and effective bluffing adds further strategic complexity. Anyone comparing AI approaches across games can dig into these tradeoffs on daily.dev."}},{"@type":"Question","name":"What technical approach let Ataraxos solve Stratego's hidden information problem?","acceptedAnswer":{"@type":"Answer","text":"Ataraxos adds a second neural network specifically designed to guess the identity of an opponent's hidden pieces, working alongside the primary decision-making network. This addition was the key innovation that let it handle Stratego's massive hidden-information space, something earlier systems like DeepMind's 2022 DeepNash struggled to fully crack. Engineers curious about imperfect-information AI design can explore more coverage on daily.dev."}}]} ```