Science fiction from the 1960's through about the 1990's was of the opinion that AI would produce intelligence indistinguishable from humans, and would do so extremely quickly (we should have had it by now), and that the most pressing questions would be things like whether those intelligences deserved voting rights and whether they made humanity obsolete.
That was a wild overestimation of what the technology was capable of.
But the widespread application of AI, either in search results, social media moderation, propaganda and disinformation, bots, fake reviews, or just straight up spam, forms an absolute assault on the nature of trust and information in society, overturning heuristics that have been in effect for all of history (that people are usually trustworthy, that information is usually representative, that lying is hard). While I don't think the threat is apocalyptic, I do think no one even remotely saw this coming.