It doesn't strike me as AI on close inspection - it looks like a very cheap CG render somebody slapped together in an afternoon in Blender. Frankly, the new Veo3 release would probably be able to make something that looks a lot better than this overall. But at the same time there are aspects like the old graphics assets on the wall (HHN32 poster) that it would be impossible for a genAI program to render with fidelity in a single-shot video generation, and that it would be a pain in the ass to paste in after the fact over a generated video, but that would be exceedingly easy to toss in there in a Blender environment.