I've noticed a new trend emerging among AI spammers, and that is to prompt the bot to generate only lowercase text.
One other thing to note is that this person posts a lot in Turkish, which could theoretically be an excuse to use AI for proof-reading of translating posts, but then telling it to decapitalize everything suggests AI is doing the heavy lifting.
Good catch! From what I've noticed, GPTZero has advanced quite a lot and is able to detect such modifications on the AI posts. It also had a paraphraser feature in beta, that detects if a post shows signs of being "humanized". The first post that I checked, even without fixing the capitalization, appears 100% AI Generated.