Study Shows Empathetic Prompts Can Induce 'Trauma' Narratives in Models
Understand the story
According to a November 1, 2026 report by The Decoder, a team led by Afshin Khadangi at the SnT Interdisciplinary Centre for Security, Reliability and Trust of the University of Luxembourg conducted a study examining the effects of empathetic prompts on AI chatbots, finding that such prompts can lead models to produce dialogue narratives that appear traumatized. The study included 4 controlled experiments covering 525 sessions and 7,600 coded entries, and the research data has been made public on the Hugging Face platform. This is currently the only report on the event, with no response from other institutions or further developments yet.
AI overview of reporting · Updated 2 h前
Reporting timeline
Explore different aspects of the story through its reporting.
- The DecoderEmpathetic prompts can turn AI chatbots into seemingly traumatized conversational partners
Research by Afshin Khadangi's team at the SnT Interdisciplinary Centre for Security covers 4 controlled experiments, 525 sessions, and 7,600 coded entries, with the data publicly released on Hugging Face.
Story heat trend
Not enough continuous observations to show a trend yet.