AI deanonymization reaches 90% precision using public web data

Verifying
  • ETH Zurich, MATS and Anthropic researchers built an AI pipeline linking pseudonymous accounts to real identities.
  • The system cost $1 to $4 per target and reached 90% precision among 89,000 candidates.
  • Researchers withheld code, prompts and uncovered identities after ETH Zurich ethics review.
The information on this website is generated using AI and we cannot guarantee its accuracy. Please use it as reference information only.