points
Why would you assume this when the more likely reasonable is that the 'content creators' are just pasting LLM output?