AI de-anonymisation risk grows as language models link anonymous social profiles
TL;DR
A Guardian report summarizes a study showing large language models can de-anonymize anonymous social media accounts by cross-referencing publicly available details, raising privacy and security concerns as attackers or governments could use AI to link identities; the researchers urge stricter data access controls and better anonymization, noting AI tools aren’t foolproof and misattribution is possible.
- AI allows hackers to identify anonymous social media accounts, study finds The Guardian
- LLMs can unmask pseudonymous users at scale with surprising accuracy Ars Technica
- AI Can Mass-Unmask Pseudonymous Accounts, Research Paper Finds Futurism
- How AI could end online anonymity Tech Xplore
- AI Risks Anonymity: Hackers & Surveillance Using LLMs Like ChatGPT National Today
Want the full story? Read the original reporting
Read on The Guardian