Artists have to submit the recording, artwork and metadata to create Art Tracks,” a YouTube spokesperson wrote in an email statement to Bellingcat.
So they say “auto generated” because the submitter submits an audio track, an image and metadata, and YouTube sticks them together for the user.
There’s no CC track included, no parsing of the Hindi lyrics, no generation or parsing of the images. So at that point, YouTube is really assembling the video track, not generating it.
And Bellingcat is attempting to capitalize on the phrasing to concern troll.