Why Is a $3 Billion MLB Team Insisting This Video Isn't AI?

6 hours ago 4

The Toronto Blue Jays posted an animated short on Tuesday and spent the afternoon telling fans it was not made with AI. Pull the video down and step through it frame by frame, and the props on a stationary wheelbarrow respell themselves four times in under a second.

The video is Spill the Tea with the Toronto Blue Jays, an 84-second cartoon in which Ernie Clement, Davis Schneider, and Mason Fluharty trade clubhouse stories. The Blue Jays are owned outright by Rogers Communications and were valued at $2.9 billion in Sportico's 2026 franchise rankings, which is the context for the reply that started the argument: a fan asking whether Rogers could not afford real animation, especially considering it is not a throwaway social clip. It is branded content, presented on screen as "Spill the Tea with Nestea," which puts an advertiser's name on the work alongside the team's. The title card bills it as "Real players. Real conversations." That line is about the stories, which are the players' own. It is the pictures that are in question.

The card shop scene, at 58.7 seconds. Some signage in this frame is clean and correct. Much of the smaller text is not, and the same words are spelled differently on different objects.

What the Team Said

Replies from the verified @BlueJays account escalated through the afternoon. To one fan, the account wrote, "Corny comment. This was made by a real animator." To another, "BREAKING NEWS: Animation existed long before the advent of AI." To a third, simply, "No, it's not." The most direct came in response to a fan asking why the team was posting AI: "NOT AI. A human animated this video. Thank you."

A fan asks whether Rogers can afford real animation. The account answers with the narrowest of the four claims, which is also the only one that would have survived the frames.

A Community Note now sits beneath one of the replies.

A flat denial.

The most direct of the four, and the one the frames contradict outright. The team has made no formal statement beyond replies from the social account.

Those four replies are not making the same claim. "This was made by a real animator" is compatible with an animator who used generative tools, and "animation existed long before the advent of AI" asserts nothing at all about this particular video, though the implications and intentions of the replies are obvious. Only the last two are falsifiable statements about how the thing was produced. Two of the posts have since been annotated with reader-added Community Notes contradicting them.

The Frame Test

You do not need a forensic tool to check a claim like this. You need the video file and the ability to export individual frames, which any free tool will do. The card shop scene runs from 57.6 to 59.9 seconds, and for most of it, the camera barely moves and a wheelbarrow full of card boxes sits still in the middle of the frame.

Here is that wheelbarrow at four points across nine tenths of a second.

One stationary prop, four frames, 0.9 seconds. Top to bottom: the Topps box, a Bowman box, the Team Bags box, the Chrome box. The Topps year reads 2024, then 2024, then 1006, then 2024, all while the Topps name weirdly morphs. Team Bags spends three frames as Team Dags. The Bowman logo reshapes in every frame.

The year on the Topps box reads 2024, then 2024, then 1006, then 2024 again. The blue box on the right says "Team Dags" for three frames and "Team Bags" for the fourth. The Bowman wordmark below it reshapes every time, drifting between Bowman and Bourman as the letters reform. The Chrome box seems to lose its "r" and get it back. Watch the way "topps" shimmers and morphs:

In traditional 2D animation, a prop is drawn once and reused as an asset. If the camera holds still and nothing in the scene interacts with a box of baseball cards, that box is byte-identical from one frame to the next, because it is literally the same artwork composited in the same place. Text on a static object cannot spontaneously change spelling, and a logo cannot redraw itself, because no one is redrawing it. Frame-to-frame instability on stationary geometry is a signature of generated video, where every frame is synthesized rather than assembled, and small details re-populate each time.

The Errors a Person Would Not Make

Separate from the shimmer, the video contains mistakes that come from a system with no knowledge of baseball rather than from sloppy drawing. A framed card on the shop wall shows Ken Griffey Jr. finishing a swing from the right-handed batter's box. Griffey batted left and threw left, and his swing is among the most recognizable in the sport's history. The poster beside it reads "ROOKIE OFHE YEAR," with the T and the word space collapsed, above a player whose jersey says "BDa" and whose hands on the bat resolve into an extra one.

Griffey swinging right-handed on the wall of a card shop, next to a poster that reads ROOKIE OFHE YEAR above a jersey lettered BDa. The hands gripping the bat on the right-hand poster do not resolve into two.

The cleanest example sits inside a single frame. The shop's hanging sign reads HOUSE OF CARDS in large, crisp, correctly spelled type. The paper bag in the wheelbarrow, three feet away in the same shot, reads HOUSE CARDS.

Both of these are in the same frame. Large text tends to survive generation intact while smaller text degrades, which is why the sign is clean and the bag is not. It's also worth noting this text style looks quite similar to many ChatGPT animation outputs seen in recent months.

The pizza scene earlier in the video has the same problem at scale. A person building a stack of fifteen pizza boxes duplicates one box and moves it fifteen times, so the lettering matches on every one. In this stack every box is wrong in a different way, and several carry illegible scribble where a word should be. Variation across duplicated objects is the opposite of what asset-based animation produces.

Fifteen pizza boxes, no two labeled alike, several carrying pseudo-text that resolves into nothing. Duplicating one drawn box would have produced fifteen identical labels.

The Denial Is the Story

The video on its own is a minor thing. It is a mediocre 84-second cartoon that would have been forgotten by Thursday, and plenty of clubs put out worse. What turned it into a story is that the account argued back, four times, with rising confidence, on a platform that timestamps everything you post and lets strangers staple a correction to it.

The likeliest explanation is not that anyone sat down and decided to deceive Blue Jays fans. A studio or a freelancer used generative tools somewhere in the pipeline, delivered a finished file, and whoever was running the account answered an angry reply with what they sincerely believed to be true. "A human animated this video" feels accurate if you hired a human animator and never thought to ask what was in their toolchain. That is a more uncomfortable position than lying, in a way, because it means a club owned by a national telecom did not know what it had bought and put an advertiser's name on it anyway. Of course, that's not the only possibility.

The asymmetry is what should worry anyone commissioning this kind of work. A claim about production method used to be effectively unfalsifiable from outside, because you had no access to the working files and no standing to interrogate a studio about its process. That is over. A generated video carries the evidence in its own frames, and extracting them costs one download and a a bit of time.

"We are checking with the studio" costs nothing, takes an hour, and cannot be disproved by a wheelbarrow that changes its mind several times over a couple seconds.

Frames captured from the Blue Jays' published video for the purpose of criticism and commentary. Social media screenshots via X.

Read Entire Article