They were super bad (like, global leaders in innovating new levels of bad) before LLMs took off, not that controversial or difficult of a distinction.
Completely agree with the commenter who replied to you, and suddenly saying "it's mostly AI" at the end frankly just lets them off the hook. Cuz they were doing this the old fashioned human way for a long time, and it bears remembering. Paying workers in other countries to traumatize themselves horrifically in order to classify and censor the worst of the worst.