So are you saying that anyone with native fluency in English but who is not from the US can't tell the difference between human writing and unpolished AI copy-paste? I don't agree. Given that US-based LLM models tend to default their output to American English, its arguably much easier for "the rest of us" to spot the "US" language patterns...