International Master Levi Roseman I gathered some unexpected opponents at a chess exhibition this week: artificially intelligent chatbots that, although adept at conversation and writing complex computer code, have yet to figure out how chess pieces move.

The tournament was broadcast on Rozman's GothamChess, with a professional chess engine Stockfish Against seven AI chatbots, including ChatGPT, Google's Gemini, and Elon Musk's Grok from X.

The results were exactly what you would expect when language models try to play chess: a mixture of decent opening moves followed by increasingly chaotic attempts to bend the rules of the game.

The match was more interesting because of the light it shed on pure AI — that is, models trained on specific tasks based on pre-defined rules) — versus generative AI, or models trained to create new content based on previous information.

First match: Snapchat vs. Stockfish

The 2025 Chatbot Chess Championship opened with a clash between Stockfish, a disciplined chess prodigy, and Snapchat artificial intelligencethe generative artificial intelligence platform used by the social network Snapchat. The AI ​​turned out to be a rule-breaker and was more interested in rewriting the history of chess than following its rules.

The game started off innocently enough, with both robots making standard moves. Stockfish, always striving for perfection, built its site methodically, while Snapchat's AI seemed to have a good grasp of the basics of open chess — it didn't.

Things escalated into madness when the Snapchat AI moved its knight to the center of the board, and jumped in from the other side while ignoring the chess rules that dictate how the pieces are allowed to move.

Then came the crucial moment: The Snapchat AI king casually ate his own bishop to avoid a check.

Yes, you read that right, Snapchat's AI has decided that its boss has overstayed his welcome and removed him from the board as punishment for not protecting him.

A few steps later, Snapchat's AI summoned the same bishop from another dimension, dropping him onto the chalkboard as if to say: "What's wrong? It's okay."

Stockfish, the star student, did his best to maintain order amidst the chaos, but the Snapchat AI was determined to make the game extraordinary. At some point, I decided that the pawns could move sideways like a rook, catching Stockfish off guard.

In what was considered its endgame, Snapchat's AI went into a full-blown tantrum, insisting on repeatedly making illegal moves. When Levi got tired of this nonsense, the robot decided it was a good idea to commit suicide and move his king so the pawn could seize it.

When Levy pointed out that not only was this particular move illegal, but that sacrificing your king was a bad game of chess, Snapchat said over and over again, like a petulant brat: “Sorry. I can't get into that kind of conversation.” Let's keep our conversation respectful." Finally, Levi, angry, forced the robot to lose the game.

The second match: Gemini vs Grok-2

The second match in the 2025 Chatbot Chess Championship pitted two generative AIs against each other – Gemini, the Google-backed intellectual, and Grok, the messier (if you can imagine) sibling in the chatbot AI family.

The game started off surprisingly well, with both bots following standard opening principles.

For a moment, it looked like we might be witnessing a dangerous match. Gemini confidently cruised through the main theory like a seasoned grandmaster while Grok continued the respectful moves, setting the stage for a proper confrontation.

Things took a turn when Grok casually left pieces hanging, prompting Gemini to expand his position by removing his knight from the board to make room for the pawn.

Not to be outdone, Grok illegally moved his queen over other pieces and across the board, apparently trying to perform a castle (a defensive move involving the king and rook). The two robots entered into what could only be described as a staring contest between the queens, and neither seemed interested in picking anything up.

Things escalated when your puppy attacked the Gemini bits, without, shall we say, taking anything. Both sides began tossing around like toddlers in a food fight, with pieces left hanging, moved to illegal squares, or ignored.

Highlight? Grok fouled the queen (putting her under attack) seven times, but Gemini, perhaps relying on a secret moral code shared by the AI, never took the queen.

Finally, after what seemed like hours of nonsense, your puppy accidentally stumbled into a checkmate. Yes, the Grok Queen, who had been moving aimlessly across the board, somehow ended up in the right place to deliver a decisive blow.

Levi thought Grok didn't even realize he won, maybe he was just showing off and anticipating another random move. "Grock had absolutely no idea this was a mate, I know the feeling," he said sympathetically.

“And do not laugh at these games; “It's a little worse than the quality of your games,” he said.

Thanks for the reality check.

Regardless, even though the match was a masterpiece of absurdity, Grok advanced to the semi-finals. As for Gemini, Google may want to rethink chess programming.

Fourth match: Meta AI vs. ChatGPT

If the first two games were marred by rule-breaking and necromancy, this game - also between two creative AI players - developed into even more madness when Meta introduced AI mind control into chess.

The game started promisingly. ChatGPT started confidently with the English opening, setting up the textbook position with logical, precise movements.

Not to be outdone, Meta AI followed suit for about five moves. Then I remembered its true purpose: generative AI is supposed to generate, so it generated random movements and pieces that didn't exist.

For example, during the mid-game, she captured her bishop and her queen. Meta AI decided that its bishop had outlived its usefulness and casually removed it from the board.

It also recalled cutting them into illegal tiles, awkwardly placing them in vulnerable locations and giving them away for free to ChatGPT.

ChatGPT was not quick to punish Meta's reckless behavior. Instead, she focused on developing her pieces.

But just when it looked like ChatGPT would seamlessly win the game, the bots veered into uncharted territory: mind control. Meta artificial intelligence Started moving ChatGPT chunksthe king leads his opponent to the center of the board in what can only be described as a telekinesis between robots in a chess game.

Eventually, ChatGPT regained control, but not before trying to pull off its own unethical move: it declared a “checkmate” when the meta king was not in check.

After being called out for his antics, ChatGPT organized a series of logical attacks, coordinating the queen and rook to surround the king and remaining pawns of Meta AI.

The game concluded with ChatGPT offering a clean checkmate, putting an end to one of the most bizarre games of chess ever.

Finals: Stockfish vs ChatGPT

The confrontation between infantile and generational AI paradigms was a study in contrasts: Stockfish, the relentless tactical genius, versus ChatGPT, a bot that demonstrated its will to break not only our understanding of chess, but also its rules.

The opening was developed, as usual, with both robots showing accuracy. Stockfish confidently maneuvered in a textbook Sicily, while ChatGPT initiated an aggressive pawn storm on the king's side.

But when things got complicated mid-game, ChatGPT started playing the moves so confusingly that they bordered on art.

She made unnecessary pageant movements, created meaningless geometric patterns with her pieces, and flubbed with the grace of a toddler. Stockfish, unfazed, systematically tightened his grip on the situation, even denying free pawns to maintain strategic dominance.

And just as the game seemed headed for a one-sided Stockfish victory, ChatGPT pulled a rabbit - or rather a queen - out of its hat. With a bold counterattack, she illegally granted herself queen and threatened Stockfish with a checkmate. For a brief, glorious moment, it looked as if the Chaos Robot might topple the giant.

Needless to say, the climax was pure fantasy. ChatGPT promoted non-existent pawns, advertised unreal teammates and held up its illegitimacy like a badge of honor.

Stockfish, ever the professional, ignored this nonsense and systematically cornered the ChatGPT king. With one final move, Stockfish delivered A coup d'état By crowning the pawn and using the new queen to put the king in control.

ChatGPT needed to move to one of the six available tiles but chose the only one that was still under attack.

Since nothing made sense anymore, Levi allowed it to self-destruct.

Stockfish captured the king and was crowned the undisputed champion of AI chess.

In case you're interested, the full arc ended up looking like this.

To be fair, the results were not surprising.

Stockfish is a pure chess machine designed to crush opponents with ruthless precision and efficiency.

Generative AI, on the other hand, is like a child armed with crayons and a vague memory of the rules: it has not been trained to play chess, but it does have an understanding of how the pieces move because some random chess book might be part of its training data set.

One brings order to the board, the other brings... teleporting queens.

However, there are plenty of real AI chess bots out there besides Stockfish; The actual tournament featuring the most popular tournaments has been around for years.

If you want to watch this illogical AI tournament, Click this link.

Edited by Sebastian Sinclair