Those can be useful, but not really against LLM content. Though neither do I think an LLM-based filter could actually reliably detect LLM-based content.