AI can link fake online names to real identities in minutes for just a few dollars

2026-03-01 · Source: The Decoder · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Cybersecurity & Data Privacy · Depth: Intermediate, medium

Summary

Commercially available AI models can de-anonymize pseudonymous internet users in minutes for $1-4 per profile, according to a study by researchers from ETH Zurich and Anthropic. This automated process, unlike earlier methods, works directly with unstructured natural language from forums and comment sections. In an experiment, an AI agent correctly linked approximately two-thirds of 338 Hacker News profiles to real identities, with a false positive rate of about ten percent. The system extracts a detailed profile from posts, compares it against candidate databases, and then uses a more powerful model to verify the most likely matches. The effectiveness increases with the volume of user-generated content, with nearly half of users with ten or more shared movie titles identified in Reddit tests.

Key takeaway

For CTOs and VPs of Engineering evaluating privacy risks, this research indicates that traditional assumptions about online anonymity for pseudonymous accounts are obsolete. Your teams should assume that any persistent online pseudonym can be linked to a real identity for minimal cost and effort. You must re-evaluate data handling policies and consider implementing stricter access controls or exploring advanced detection mechanisms for automated scraping to mitigate this heightened risk.

Key insights

AI models can de-anonymize online pseudonyms from natural language posts quickly and cheaply.

Principles

• Unstructured text is now de-anonymizable.
• More data increases identification accuracy.

Method

The de-anonymization pipeline involves four stages: profile distillation from posts, candidate matching via similarity search, individual candidate verification by a powerful model, and confidence-based decision making.

In practice

• Assume persistent usernames are linkable.
• Restrict access to user data for defense.

Topics

• Online De-anonymization
• Large Language Models
• Pseudonymity
• Natural Language Processing
• AI Agents

Best for: CTO, VP of Engineering/Data, Director of AI/ML, AI Researcher, AI Ethicist, Policy Maker

Related on AIssential

• AI allows hackers to identify anonymous social media accounts, study finds — LLMs significantly simplify de-anonymization of social media accounts, posing a new privacy risk.
• Large-scale online deanonymization with LLMs (Simon Lermen, Daniel Paleka et al., — LLMs automate and scale online deanonymization, rendering traditional pseudonymous privacy ineffective against low-cost, large-scale attacks.
• LinkedIn user hides AI prompt injection in bio to force recruitment spam to be sent in Olde English prose — bots also also manipulated to address user as ‘My Lord’ — AI agents are vulnerable to indirect prompt injection from untrusted external data sources.
• The hidden costs of cheap AI APIs — Cheap AI API proxies pose severe security risks and often deliver inferior models, making their perceived savings illusory.
• AI Has Turned Internet Fraud into a Transnational Industry — AI has transformed internet fraud into a sophisticated transnational industry, enabling highly personalized and scalable attacks.
• Reddit will require "fishy" accounts to verify they are run by a human — Reddit is implementing human verification and bot labeling to preserve authentic human interaction amidst rising AI bot traffic.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by The Decoder.