New research reveals Large Language Models can now identify pseudonymous online users at scale, outperforming traditional deanonymization methods and raising significant privacy concerns.
