BUILDER'S SANDBOX
Build This Paper
Use an AI coding agent to implement this research.
OpenAI CodexAI Agent
Lightweight coding agent in your terminal.
Claude CodeAI Agent
Agentic coding tool for terminal workflows.
AntiGravity IDEScaffolding
AI agent mindset installer and workflow scaffolder.
CursorIDE
AI-first code editor built on VS Code.
VS CodeIDE
Free, open-source editor by Microsoft.
Recommended Stack
Startup Essentials
MVP Investment
$10K - $13K
6-10 weeksEngineering
$8,000Cloud Hosting
$240SaaS Stack
$800Domain & Legal
$5006mo ROI
2-4x
3yr ROI
10-20x
Lightweight AI tools can reach profitability quickly. At $500/mo average contract, 20 customers = $10K MRR by 6mo, 200+ by 3yr.
Talent Scout
S
Simon Lermen
MATS
D
Daniel Paleka
ETH Zurich
J
Joshua Swanson
ETH Zurich
M
Michael Aerni
ETH Zurich
Find Similar Experts
Privacy experts on LinkedIn & GitHub
Founder's Pitch
"Use LLMs to deanonymize online profiles at scale with high precision."
Commercial Viability Breakdown
0-10 scaleHigh Potential
2/4 signals
Quick Build
4/4 signals
Series A Potential
2/4 signals
Sources used for this analysis
arXiv Paper
Full-text PDF analysis of the research paper
GitHub Repository
Code availability, stars, and contributor activity
Citation Network
Semantic Scholar citations and co-citation patterns
Community Predictions
Crowd-sourced unicorn probability assessments
Analysis model: GPT-4o · Last scored: 2/18/2026
🔭 Research Neighborhood
Generating constellation...
~3-8 seconds
Why It Matters
This research highlights a critical vulnerability in online anonymity, showcasing how LLMs can deftly deanonymize pseudonymous accounts using unstructured text. Without addressing such vulnerabilities, individuals’ privacy online is significantly compromised, granting unauthorized access to their real identities and associated sensitive information. Organizations and platforms must revise privacy measures to safeguard against these scalable, AI-driven deanonymization attacks.
Product Angle
This can be transformed into a software service that audits the vulnerability of online systems to deanonymization by AI, providing insights and solutions for enhanced privacy and security.
Disruption
This approach could disrupt the current market for privacy auditing and personal data security solutions by providing an automated, AI-driven method of assessing deanonymization risks far more effectively than manual efforts.
Product Opportunity
Given the growing concern about privacy in the digital age, platforms and privacy-focused applications could pay to ensure their users are protected from such AI-driven deanonymization attacks, making it a vital tool for any service handling user data.
Use Case Idea
Develop a privacy auditing tool that uses LLMs to simulate potential deanonymization of individuals to test and enhance privacy measures for online platforms.
Science
The paper proposes a novel approach where Large Language Models (LLMs) autonomously deanonymize online profiles by analyzing unstructured text data across diverse platforms. The process harnesses LLMs to extract identity-relevant features from text, create embeddings for these features, and cross-reference profiles to confidently establish identity matches. This method significantly outperforms traditional deanonymization approaches that relied heavily on structured data.
Method & Eval
The tool was evaluated in various settings with datasets linking Hacker News to LinkedIn profiles, Reddit discussions, and multi-profile anonymization, achieving superior recall and precision than traditional approaches — up to 68% recall and 90% precision compared to near zero for baseline methods.
Caveats
Although effective, this method currently depends on available public data, which could limit its scope if data access is restricted. There's also the ethical concern of using AI for identity matching without consent, which must be navigated responsibly.
Author Intelligence
Simon Lermen
MATS
Daniel Paleka
ETH Zurich
Joshua Swanson
ETH Zurich
Michael Aerni
ETH Zurich
Nicholas Carlini
Anthropic
Florian Tramèr
ETH Zurich