…researchers show that modern large language models (LLMs) can re-identify people behind pseudonymous online accounts at a scale and accuracy that far surpass previous techniques.

https://www.zerohedge.com/markets/ai-can-now-unmask-anonymous-internet-users-new-study-finds

AND

Large-scale online deanonymization with LLMs

https://arxiv.org/pdf/2602.16800