When Baseball Banned AI: What MLB’s Dugout Crackdown Means for Every Industry
Major League Baseball banned AI-powered apps from dugout iPads after a third of teams used them for in-game decisions. It’s a preview of the boundaries every industry will need to draw.
Major League Baseball has drawn a line in the dirt. After nearly a decade of teams using league-issued iPads in the dugout, the league has banned custom AI-powered apps that were quietly influencing in-game decisions. According to reporting from The Verge, up to a third of MLB teams had been using generative AI tools on their dugout tablets to recommend pitch calls, player substitutions, and other strategic decisions traditionally made by coaches and players.
The iPad in the Dugout
Since 2016, Apple and MLB have had a partnership that put iPad Pros in every dugout across the league. The devices were originally intended for video replay review and accessing player statistics — a digital upgrade over the bulky binders managers used to carry. But over the years, some teams went further. Custom apps tapping into generative AI began appearing on those tablets, offering real-time recommendations for:
- Pitch selection and sequencing based on batter tendencies
- Defensive positioning shifts based on game situation
- Pinch-hitter and relief pitcher matchups
- In-game substitution patterns optimized by win probability
The league has now put a stop to it. Teams can no longer install custom apps that leverage generative AI for these purposes. The iPads remain, but their software has been reined in.
Why the Ban Matters
This is not just a baseball story. It is a preview of the tension that will play out across every industry where AI threatens to replace human judgment with algorithmic optimization.
Baseball has always been a sport that embraces data. Billy Beane and the Oakland Athletics popularized sabermetrics in the early 2000s, a story told in the book and film Moneyball. Every front office now employs armies of analysts. But there is a critical difference between analytics and AI. Analytics inform decisions. AI makes them. When a manager pulls up a dashboard and sees that a left-handed reliever has a .310 OBP against right-handed hitters in night games, that is a tool. When an AI app says "bring in Smith for the next batter" based on a model no human can fully explain, that is something else entirely.
MLB seems to recognize this distinction. The spirit of the game — the chess match between manager and opponent, the gut feel of when to pull a pitcher — is something the league wants to protect. And they are right to do so before the slide becomes irreversible.
The Broader Pattern: Industries Drawing Lines
MLB is not the only institution pushing back against AI this month. Several recent stories highlight a growing willingness to set boundaries:
- San Francisco city attorney David Chiu sent cease-and-desist letters to Apple and Google demanding the removal of 13 AI "nudify" apps that generate non-consensual explicit images. Apple confirmed it removed the apps and is terminating developer accounts.
- New York City mayor Zohran Mamdani issued new housing policies requiring landlords to disclose when rental listings have been altered by AI, cracking down on misleading real estate photos.
- Meta Oversight Board testing found that leading LLMs from Anthropic, DeepSeek, Google, Meta, and OpenAI are significantly less likely to criticize governments with restrictive speech policies — raising questions about model bias and corporate influence.
The common thread is not anti-AI sentiment. It is a recognition that AI without guardrails degrades the things people care about — fair competition, consent, transparency, and the human elements that make activities meaningful.
The Competitive Problem
One of the thornier issues MLB faces is competitive balance. If a third of teams were using AI tools and two-thirds were not, there is a fairness question. Were the AI-assisted teams gaining an edge? The league has not disclosed which teams were using the apps, but the implication is clear: some teams were getting algorithmic help that others did not have.
This mirrors concerns in other domains. In education, AI detection tools like Pangram are being trusted to identify AI-written text, but the playing field is uneven when some students use AI and others do not. In real estate, AI-altered listing photos give some sellers an unfair advantage. The competitive integrity question is universal.
What Baseball Gets Right
MLB is making the right call by acting before AI in the dugout becomes entrenched. Once a competitive advantage takes hold in professional sports, it is nearly impossible to remove without disrupting the entire ecosystem. The league still allows data analysis, video review, and statistical lookup on the iPads. What it has done is prevent the devices from becoming autonomous decision-makers.
There is a lesson here for every organization experimenting with AI. The question is not just "Can AI do this?" but "Should AI do this, and where does it stop?" Baseball has answered that question for its sport. Other industries will need to answer it for theirs.
The Road Ahead
The MLB ban is a small but significant moment in the broader AI regulation landscape. While Congress debates sweeping legislation and the EU enforces its AI Act, individual organizations are making practical decisions about where AI belongs and where it does not. These micro-decisions may matter more than sweeping policy in the short term.
For baseball, the game will go on. Managers will still pull pitchers too early or too late. Coaches will still argue with umpires. And fans will still second-guess every decision from their seats. The difference is that those decisions will be made by humans again — with all the brilliance and mistakes that entails. That is exactly how it should be.
Related Posts
Varkos: The AI Gaming Companion That Actually Plays With You
A developer built an AI dog companion for Skyrim that understands voice commands, executes multi-step plans, and evolves its personality over time — all running on local hardware with sub-500ms latency.
Why Your Local LLM Feels Dumber Than It Is: The Hidden Quality Gap
Your local LLM is not broken. Quantization, weak system prompts, and basic inference engines silently degrade quality. Here is what to fix.
AI Blindness: When Your Brain Learns to Stop Reading AI-Generated Content
A growing number of people report their brains automatically filtering out AI-generated text, like banner blindness for LLM output. This phenomenon reveals something deeper about trust, attention, and the future of human-AI interaction.