On 6 October 2026, the Neue Zürcher Zeitung’s daily news podcast «NZZ Akzent» devoted an episode to our study on agentic AI and mobility data: «Massenüberwachung per KI: Selbst die Forscher sind überrascht, wie leicht das geht» (in German).
In the episode, NZZ technology editor Gioia da Silva, who interviewed Oscar Thees for her earlier NZZ article, explains how the study came about and what it means; the episode also features clips of Oscar Thees himself. It started with a data broker offering location data on more than 800,000 smartphones in Switzerland free of charge. Analysing such data was out of the question for ethical reasons, so the team instead tested on simulated data whether today’s AI agents can identify people from location traces and learn about their homes, workplaces and daily routines. The result surprised the researchers themselves: it was far easier than expected.
All coverage of the study, including the NZZ commentary, is collected in our post on the NZZ article.
- Podcast: NZZ Akzent, 6 October 2026
- Paper: Agentic AI-Powered Re-identification, PSD 2026, Springer LNCS
Cover illustration: Hans-Jörg Walter / NZZaS.