MLB Cracks Down on Dugout AI, and the Mets Land in the Crosshairs

Major League Baseball has begun restricting iPad use in dugouts, a move aimed at preventing teams from running artificial-intelligence tools on the tablets during games, according to a 17 July 2026 ESPN report. The directive lands at a sensitive moment for the New York Mets, who were singled out in reporting on in-game AI usage that surfaced this week.

At Citi Field, interim manager Andy Green told reporters on 18 July 2026 that the club is fully compliant with the league on the matter. The exchange, captured by ESPN, put the Mets on record as the team most visibly named in the early coverage of the new restriction, even as the rule itself applies league-wide.

What the league actually changed

The mechanics are narrow but telling. Per the 17 July ESPN report, MLB is disallowing AI usage on iPads inside dugouts, the same Apple-tablet setup teams have used for years to review hitter tendencies, bullpen matchups, and spray charts. The league did not ban iPads outright; it curtailed what those iPads are permitted to do. The distinction matters. Video review, scouting databases, and statistical look-ups remain available; algorithmic decision support, the kind of tool that would ingest a pitcher-catcher sequence and flag a recommended action in real time, does not.

The restriction reads as a pre-emptive move. Rather than waiting for a public dispute over an AI-assisted decision, MLB is signalling that the human manager still owns the bench. That framing protects the league's labour architecture. Managers are members of a union-negotiated hierarchy with on-field authority. If a tablet began nudging pitching changes or pinch-hitting decisions, the question of who is calling the game becomes a contractual, not just a tactical, one.

Why the Mets got named

ESPN's reporting identified the Mets by name as the club whose in-game AI usage drew attention. Green, taking questions on 18 July 2026, said the organisation is "fully compliant" with MLB on the matter, the strongest version of a denial a manager can offer without disclosing proprietary workflows. The team's posture is straightforward: whatever was being run on the iPads was within the rules that existed at the time, and the new rules have been absorbed.

Two things are worth keeping in mind. First, being named in a leak about a behaviour the league has now prohibited does not, on its own, imply wrongdoing under the prior rulebook. The club is the only one on the public record so far, and that scarcity of names invites speculation the team would rather not manage. Second, the league's enforcement style matters. By issuing a blanket prohibition rather than pursuing a team-specific penalty, MLB keeps the dispute administrative rather than disciplinary, which is what most clubs in this position prefer.

What AI in a dugout would have looked like

The reporting describes AI assistance for in-game decisions. In practice, that covers a wide spectrum. At the soft end, a tablet running a model that ingests catcher framing data to suggest a pitch sequence is a tool that sharpens a coach's existing instinct. At the hard end, a system that triggers a bench-coach prompt when a pitcher's spin-rate drops below a threshold effectively pre-authorises a mound visit. MLB's restriction targets both. The line drawn is the line between recommendation and execution, and the league has put the line at the tablet.

This is not a new tension. Pitch-framing analysis, Statcast metrics, and shift-geometry models have shaped in-game calls for a decade. What is new is the integration layer: a single interface that fuses video, biomechanics, and probability outputs into a single decision prompt. Once that layer exists, the question is no longer whether AI is influencing the game; it is who carries the liability when it does.

The larger pattern across the league

MLB is moving earlier and faster on AI governance than most professional leagues have on comparable issues. The dugout iPad is the visible frontier because the in-game setting compresses every rule question into a few seconds of action. A model recommending a bunt in the seventh inning of a tie game is a different artefact from the same model generating off-day scouting notes, and the league has chosen to police the former while leaving the latter untouched.

For the Mets, the next test is procedural. If MLB formalises the iPad rule into the operations manual, every team will need an auditable workflow showing which applications are loaded on which device, and who has access. The Mets, having been named once, will live under that audit longer than clubs that were not. That is the cost of being the example, even when the answer to the question is, as Green put it, full compliance.

The reporting on this story is still thin. The 17 July ESPN dispatch does not name other clubs, describe specific products in use, or quote any league memo, and the 18 July Mets response is a manager's statement rather than a documentary release. What remains to be seen is whether the league publishes the rule in writing, whether any team is sanctioned for past usage, and whether the iPad workflow becomes a routinely disclosed compliance item at the start of each season.

Desk note: Monexus reported the league directive and the Mets' on-record response, and declined to speculate on which other organisations may have been using AI tools in dugouts, as the available reporting does not name them.