[ozbonus]

In the back of mind I knew it wasn't so, but I had been holding onto the belief that surely I could discern between human and bot, and that bots weren't a real issue where I spent my time anyway. But no. We're at a point where any anonymous public comment is possibly an impersonation. And eventually that "possibly" will have to replaced with "most likely".

I don't know what the solution is or if there even is one.

[int_19h]

There isn't. Not only LLMs are good enough to fool humans like this, but they have been that for quite a while now with the right prompting. A large number of readily available open weights models can do this, so even if large providers were to crack down on this kind of use, it's still easy to run the model locally to generate such content. The cat is well and truly out of the bag.