AI Was Helping Call Pitches From the Dugout. MLB Just Shut the Door

Baseball has spent years becoming one of the world’s most data-driven sports. Now Major League Baseball has decided that one form of digital assistance went too far.

MLB restricted custom programs on league-issued dugout iPads at the start of the second half of the 2026 season. The tablets can still show approved video and league data, but teams can no longer use custom tabs that had evolved into tools for recommending substitutions, pitch selection and other decisions during a game.

How AI reached the dugout

Baseball clubs already collect enormous amounts of information about pitch movement, batter tendencies, defensive positioning and player fatigue. Machine learning can search those patterns faster than a human analyst and estimate which pitch or matchup has the highest probability of success.

Before a game, this kind of analysis is routine. The controversy begins when a recommendation engine sits only a few taps away from coaches and players while the game is happening.

According to the Associated Press, MLB executive Morgan Sword sent teams a memo on June 11 explaining that custom iPad tabs were going beyond their original purpose. The change took effect after the All-Star break, giving clubs time to adjust.

Why the Mets became part of the story

Former major-league pitcher Adam Ottavino said on a YouTube livestream that the New York Mets were one of the teams using an expensive AI system and that their activity helped prompt the crackdown. He described technology capable of assisting with pitch calls and other decisions.

That is an allegation from a former player, not an official finding against the Mets. The club did not publicly respond to the claim in the initial reports. MLB said its review found teams compliant with the regulations that were in effect, and no club was punished.

The distinction matters. The league is closing a loophole and defining a new boundary, not announcing that it caught a team cheating.

The line between information and instruction

An iPad showing a pitcher’s previous at-bats is a library. An application that processes the current situation and tells a catcher which pitch to request acts more like a coach. Both rely on data, but the second system moves from evidence to live instruction.

That creates three concerns. First, wealthy clubs can buy better models, more computing power and larger analytics teams. Second, a real-time recommendation system may reduce the role of human judgment that fans consider part of the contest. Third, teams might gradually add capabilities faster than league rules can evaluate them.

There is also a competitive argument in AI’s favor. Better analysis can improve player preparation, reduce avoidable mistakes and reveal strategies that humans miss. If every club had equal access under transparent rules, some fans might see it as the next stage of the sport rather than an intrusion.

Baseball’s complicated history with screens

MLB first tested iPads in dugouts in 2015 and expanded their use in 2016. Access to in-game video was later removed during the fallout from the Houston Astros sign-stealing scandal, then returned in a controlled form in 2021.

The new AI restriction continues that cautious approach. Baseball accepts technology when it improves review and preparation, but becomes more defensive when a screen can shape the next live decision.

The timing is especially interesting because MLB is also using an automated ball-strike challenge system. Technology is therefore welcome when it helps settle a defined officiating question, while AI-generated strategy remains much harder to regulate.

Can the rule actually be enforced?

Removing custom tabs from official iPads closes the clearest path, but it does not eliminate algorithmic strategy. Clubs can still use AI before games, print recommendations, brief players and encode model outputs into conventional scouting plans.

The difficult regulatory question is not whether teams may analyze data. It is how close an automated recommendation can get to the live action, and how quickly it can respond, before it becomes prohibited assistance.

MLB will eventually need rules that define permitted data, update frequency, device access, audit trails and penalties. Simply banning one interface may not be enough as AI tools become smaller, faster and easier to hide inside ordinary software.

A preview of arguments coming to every sport

The dugout dispute is bigger than baseball. Coaches in football, basketball and motorsport already depend on simulation and predictive analytics. As AI starts issuing recommendations in real time, leagues must decide which skills belong to athletes and coaches, and which decisions may be delegated to machines.

MLB’s answer, for now, is that the tablet may show the evidence, but it should not become the manager.