LLM-Based System Achieves 68% Recall at 90% Precision for Online User Deanonymization

Researchers demonstrate that large language models can effectively deanonymize online users by analyzing their writing style and content across platfo