Their technical report (arxiv.org/abs/2402.14873) goes into it. They train a classifier on human text paired with AI-generated versions of the same documents.

Their technical report (arxiv.org/abs/2402.14873) goes into it. They train a classifier on human text paired with AI-generated versions of the same documents. Dataset scale is comparable to LLM pretraining sets.