Hackers can use AI to de-anonymise social media accounts
Researchers warn LLMs enable cheap, personalised privacy attacks, from scams to government surveillance.
AI technology behind platforms like ChatGPT is making it significantly easier for hackers to identify anonymous social media users, a new study warns. LLMs could match anonymised accounts to real identities by analysing users’ posts across platforms.
Researchers Simon Lermen and Daniel Paleka warned that AI enables cheap, highly personalised privacy attacks, urging a rethink of what counts as private online. The study highlighted risks from government surveillance to hackers exploiting public data for scams.
Experts caution that AI-driven de-anonymisation is not flawless. Errors in linking accounts could wrongly implicate individuals, while public datasets beyond social media- such as hospital or statistical records- may be exposed to unintended analysis.
Users are urged to reconsider what information they share, and platforms are encouraged to limit bulk data access and detect automated scraping.
The study underscores growing concerns about AI surveillance. While the technology cannot guarantee complete de-anonymisation, its rapid capabilities demand stronger safeguards to protect privacy online.
Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!