Research
154d agoLLMs Expose Online Identities at Scale
Large-scale deanonymization using LLMs shows that conversational models can reconstruct personal data from public posts, raising concerns about privacy worldwide. Researchers demonstrate that a single prompt can link usernames to real-world identities across platforms, exposing sensitive information.
The Signal
The technique threatens individuals, businesses, and regulators, prompting calls for stronger data-protection standards and transparency from AI developers like OpenAI and Google.