“In Europe, regulators are using Large Language Models to enforce existing hate speech laws at a scale that was previously impossible.”