Tag: llm

LLM Agents Achieve Scalable De-anonymization Across Social Platforms

Security researchers have shown that large language models can systematically de-anonymize users across social platforms like Reddit and Hacker News by inferring personal details from anonymous posts and cross-referencing web searches. The technique scales to tens of thousands of potential targets, representing a fundamental shift in the practical feasibility of mass de-anonymization.