The AIs/LLMs are getting better so it's getting harder for people in general to tell.
Accusations of using AI or "compliments" about "being as good as" AI (pro-tip for people that think they're giving a compliment when you say this, a number of artists would see it as a grave insult) seem to be a bit more common now as a result (it's not a new problem for digital artists, we've been getting "the computer is doing it all for you" since that became a thing XD), because someone happens to be actually good at what they do.
I had a sudden conspiracy theory that the LLMs are doing deliberately bad code so that you have trouble detecting LLM generated stuff which was much more entertaining than the more logical LLMs are also not brilliant at coding and have the same problems as everyone else of being able to tell what they generated XD
RE: Explorations in failure - trying to get a LLM to detect LLM generated content on HIVE