My worry is about LLM driven news. You find a lot of free news sites are just summarized lazy versions of paid news like NYT or something else. But GPT can already do just as good summaries than those people (it really isn't hard) but certainly it is a lot cheaper than paying a person literally any rate by the hour. But how does the source material that is being leeched off of then survive when it costs less than a penny for what cost the source likely thousands. AI is pretty far away from performing its own investigative reporting where it knows who to reach out to, talk to, and perform all such actions coherently and consistently.

But it also makes it easy to spam shit like this where you just write 5 articles with different sentiment and send it out. It can even be an online system, optimizing headlines for higher "engagement". But engagement is a proxy and I think we need to start actively acknowledging that the metric has been long hacked and at this point abused.