Results of the summarizing process for the arXiv paper: 2602.16800v1

In their research project titled "Large-scale online deanonymization with LLMs," Simon Lermen, Daniel Paleka, Joshua Swanson, Michael Aerni, Nicholas Carlini, and Florian Tramèr explore the use of large language models (LLMs) for performing at-scale deanonymization. They demonstrate that with full Internet access, their agent can accurately re-identify Hacker News users and Anthropic Interviewer participants using only pseudonymous online profiles and conversations. This task would typically require hours for a human investigator. The researchers further develop attacks for the closed-world setting by utilizing two databases of pseudonymous individuals containing unstructured text related to each individual. They implement a scalable attack pipeline leveraging LLMs to extract identity-relevant features, search for candidate matches through semantic embeddings, and reason over top candidates to verify matches and reduce false positives. This approach stands out from prior deanonymization work as it operates directly on raw user content across various platforms without the need for structured data or manual feature engineering. To evaluate their attacks, three datasets with known ground-truth data are constructed. The first dataset links Hacker News users to their LinkedIn profiles based on cross-platform references present in the profiles. The second dataset involves matching users across Reddit movie discussion communities while the third dataset splits a single user's Reddit history in time to create two pseudonymous profiles for matching purposes. In each scenario, the LLM-based methods significantly outperform classical baselines achieving recall rates of up to 68% at 90% precision compared to near 0% for non-LLM methods. These results highlight that the practical obscurity protecting pseudonymous users online is no longer effective and emphasize the need to reassess threat models concerning online privacy. By showcasing the capabilities of LLMs in deanonymization tasks across diverse platforms and datasets, this research underscores the importance of addressing privacy concerns in an increasingly interconnected digital landscape.

• Researchers explore using large language models (LLMs) for at-scale deanonymization
• Agent accurately re-identifies Hacker News users and Anthropic Interviewer participants using only pseudonymous online profiles and conversations
• Attacks developed for closed-world setting utilizing databases of pseudonymous individuals with unstructured text
• Scalable attack pipeline leverages LLMs to extract identity-relevant features, search for candidate matches through semantic embeddings, and reason over top candidates to verify matches and reduce false positives
• Approach operates directly on raw user content across various platforms without structured data or manual feature engineering
• Evaluation conducted on three datasets with known ground-truth data, showing LLM-based methods significantly outperform classical baselines with recall rates up to 68% at 90% precision
• Results highlight the ineffectiveness of practical obscurity protecting pseudonymous users online and emphasize the need to reassess threat models concerning online privacy

Summary- Researchers are studying how to use big language models to figure out who people are online. - An agent was able to accurately find out who some users of Hacker News and Anthropic Interviewer were by looking at their online profiles and conversations. - The attacks were made for a specific situation where they had information about people with fake names and unstructured text. - They used a method that can quickly find important information about someone, look for possible matches, and check if they are correct. - This method works on what people write online without needing organized information or manual work. Definitions- Researchers: People who study things to learn more about them. - Large language models (LLMs): Big computer programs that understand and generate human language. - Deanonymization: Finding out the real identity of someone who is trying to stay anonymous. - Pseudonymous: Using a fake name instead of your real one online. - Scalable: Able to grow or change in size easily.

Introduction In today's digital age, privacy has become a major concern for individuals who use online platforms. With the rise of social media and other online communities, people often share personal information without realizing the potential consequences. This has led to an increase in research on deanonymization techniques, which aim to uncover the identities of pseudonymous users online. A recent research project titled "Large-scale online deanonymization with LLMs" by Simon Lermen, Daniel Paleka, Joshua Swanson, Michael Aerni, Nicholas Carlini, and Florian Tramèr explores the use of large language models (LLMs) for performing at-scale deanonymization. The researchers demonstrate that with full Internet access, their agent can accurately re-identify Hacker News users and Anthropic Interviewer participants using only pseudonymous online profiles and conversations. Background The concept of deanonymization is not new; it has been studied extensively in various fields such as computer science and sociology. However, previous approaches have relied on structured data or manual feature engineering to identify individuals behind pseudonyms. This approach is time-consuming and limited in its scope. The use of large language models (LLMs) offers a more efficient and scalable solution for deanonymization tasks. LLMs are powerful natural language processing tools that can process vast amounts of text data and extract meaningful features without human intervention. Methodology To evaluate their attacks using LLMs for deanonymization purposes, the researchers constructed three datasets with known ground-truth data. The first dataset links Hacker News users to their LinkedIn profiles based on cross-platform references present in the profiles. The second dataset involves matching users across Reddit movie discussion communities while the third dataset splits a single user's Reddit history in time to create two pseudonymous profiles for matching purposes. The attack pipeline developed by the researchers leverages LLMs to extract identity-relevant features from raw user content across different platforms. These features are then used to search for candidate matches through semantic embeddings and reason over top candidates to verify matches and reduce false positives. Results The results of the study show that LLM-based methods significantly outperform classical baselines in all three datasets, achieving recall rates of up to 68% at 90% precision compared to near 0% for non-LLM methods. This highlights the effectiveness of LLMs in deanonymization tasks across diverse platforms and datasets. Implications The findings of this research have significant implications for online privacy. With the increasing use of social media and other online platforms, individuals may unknowingly share personal information that can be used to identify them. The practical obscurity protecting pseudonymous users online is no longer effective, and it is essential to reassess threat models concerning online privacy. Conclusion In conclusion, the research project "Large-scale online deanonymization with LLMs" demonstrates the power of large language models in performing at-scale deanonymization tasks across various platforms and datasets. By showcasing their capabilities, this study emphasizes the need to address privacy concerns in an increasingly interconnected digital landscape. It also calls for further research on developing effective measures to protect individuals' identities while using online platforms.