Large Language Models (LLMs) have demonstrated an alarming ability to unmask individuals online, potentially eroding anonymity on the internet. A study by ETH Zurich found that LLMs can quickly combine disparate online information to identify users, accomplishing in minutes what would take human investigators hours1. The research used anonymous bios from real profiles on HackerNews and Reddit, tasking LLM agents with scouring the internet for additional details to reveal the users' identities. While the results were varied, the implications are significant, as LLMs can automate and accelerate the process of deanonymization. This development poses a substantial risk to online anonymity, as LLMs can potentially be used to uncover sensitive information about individuals. So what matters to practitioners is that LLMs can now be leveraged to bypass traditional anonymity measures, underscoring the need for more robust online protection strategies.