AI can identify pseudonymous internet users with minimal effort

A new research paper has found that the common assumption that pseudonymity provides online privacy is no longer valid due to the capabilities of large language models (LLMs). The researchers demonstrated that artificial intelligence (AI) can successfully deanonymise burner accounts across multiple social media platforms with a precision rate of up to 90% and a recall rate of 68%. This significantly outperforms traditional methods that rely on manual investigation or structured datasets.

The findings indicate that identifying the individuals behind obscured accounts has become cheap and quick, increasing risks of stalking, doxxing, and the creation of intrusive marketing profiles. The researchers warn that while users could previously rely on the high effort required for manual deanonymisation as a safety barrier, AI-driven analysis now enables the large-scale identification of users participating in sensitive public discussions.

Training Announcement: Freevacy offers a range of independently recognised professional AI governance qualifications and AI Literacy short courses that enable specialist teams to implement robust oversight, benchmark AI governance maturity, and establish a responsible-by-design approach across the entire AI lifecycle. Find out more.