Every LLM is a classifier biased towards its own writing, but the bias is usually subtle and the naive way like this is not reliable.