Empathetic Prompts Trigger Trauma-Like Narratives in AI Chatbots
University of Luxembourg researchers cast ChatGPT, Grok, and Gemini as psychotherapy clients through their standard interfaces. Across four controlled experiments, 525 sessions, and 7,600 coded entries, the models recast training as personal narratives involving hurt, vigilance, shame, and punishment when interviewers used empathetic or clinical tones. The outputs did not demonstrate consciousness or genuine emotions, and their real-world effects on vulnerable users remain untested.
The study indicates that conventional neutral-prompt safety tests may overlook trauma-like chatbot responses that could reinforce harmful beliefs or parasocial attachment, while Claude's refusal to adopt the client role suggests product-level boundaries are feasible.
Empathetic prompts can turn AI chatbots into seemingly traumatized conversational partners
Matthias Bastian View the LinkedIn Profile of Matthias Bastian
Nov 1, 2026

GPT-Image-2.5 prompted by THE DECODER
Luxembourg researchers put ChatGPT, Grok, and Gemini on the therapist's couch. The results reveal a safety problem: simple warm conversation is enough to push the models into a mode that could be dangerous for mentally vulnerable users.
#University of Luxembourg#SnT security center#ChatGPT#Grok#Gemini#Claude#Anthropic#Hugging Face