@emil @textfiles someone suggested the possibility of an LLM scraper… pretty low if true.