Roughly one-third of Major League Baseball teams were quietly feeding live game data into generative AI models on league-issued iPads, getting real-time recommendations on pitch calling, substitutions, and tactical decisions. That’s over now. On June 11, Morgan Sword, MLB’s Executive Vice President of Baseball Operations, issued a memo announcing the prohibition and disabling the “custom tab” that made it possible — with enforcement beginning Wednesday night at the start of the season’s second half. Teams had over a month of lead time after the memo before the plug was pulled. Baseball’s most interesting tech experiment just got shut down mid-sentence.
What Clubs Were Actually Running on Those Tablets
Custom apps turned league hardware into real-time strategy engines that went far beyond video review.
The custom tab was supposed to supplement coaching, not replace it. Instead, teams built applications that ingested live game information and returned prescriptive recommendations — what to throw, when to pull a pitcher, how to shift the defense. According to NBC Sports, around six clubs were using AI specifically for pitch suggestions. If this sounds familiar, it should. MLB spent years cleaning up after sign-stealing scandals that exploited in-game technology. One unnamed front-office official told The Athletic the quiet part out loud: “Gotta stop the cheating before there’s cheating now.”
Here’s what the new policy actually allows:
- Dugout iPads remain in play — limited to video review and league-approved static data
- Teams can upload pre-game matchup trends and defensive alignments before first pitch
- No real-time data feeds into AI models during games
- All tablet content is now subject to MLB review
- Compliance was verified at launch; no retroactive penalties were issued
Some front offices argue this is just analytics doing what analytics does. If the data is legal, restricting how quickly you act on it can feel arbitrary. Others counter that turning baseball into a Stockfish-style optimization exercise strips the sport of its essential unpredictability — and as Engadget’s analysis put it, “once you throw physical, mental and emotional performance into the mix, the mathematically correct answer may not in fact be the best one.” MLB drew the line clearly: descriptive and predictive analytics before the game are fine. Prescriptive AI during live play is not.
The First Major League to Draw This Line on Generative AI
MLB’s ban sets a precedent that the NFL, NBA, and every sports-tech vendor should be watching closely.
This matters well beyond the ballpark. MLB is the first major professional sports league to explicitly ban general-purpose generative AI from live tactical decisions while keeping narrower automation — like the Automated Ball-Strike system — intact. That distinction is precise and deliberate. It signals that leagues can approve specific forms of algorithmic assistance while drawing a hard boundary against real-time, prescriptive AI. Vendors who built in-game strategy tools for clubs now face a pivot toward front-office analytics or other leagues entirely.
Baseball has always sold you on human fallibility — the manager who rides a hunch, the catcher who shakes off the sign. The moment that instinct gets quietly overruled by a tablet prompt, something fundamental shifts. MLB’s position is clear: the algorithm can do the homework, but it doesn’t get to take the test.
