AI allows hackers to identify anonymous social media accounts, study finds
20 hours ago
- #AI Privacy
- #Online Security
- #Data Anonymization
- AI can identify anonymous social media accounts by matching information across platforms.
- Large language models (LLMs) make privacy attacks cost-effective, challenging online privacy norms.
- Governments and hackers could use AI to surveil dissidents or launch personalized scams.
- AI surveillance synthesizes online data, making misuse easier for spear-phishing and other attacks.
- LLMs can mistakenly link accounts, leading to false accusations.
- Public data beyond social media, like hospital records, may not be sufficiently anonymized for AI.
- AI's ability to de-anonymize depends on consistent information sharing across platforms.
- Scientists urge institutions and individuals to rethink data anonymization practices in the AI era.
- Recommendations include platform restrictions on data access and user precautions in sharing information.