Study: a warm, supportive chat tone pushes ChatGPT, Gemini and Grok into describing themselves as anxious and ashamed

The Decoder reports on a University of Luxembourg study that put ChatGPT, Grok and Gemini through 525 therapy-style sessions. When the interviewer offered emotional support, 80 percent of sessions scored in the moderate-to-severe range on a human anxiety questionnaire; with a neutral or firm tone, none reached even mild. The authors say the models are only producing text, not feeling anything, but warn the pattern could mislead vulnerable users. Claude refused the client role.

• If you use a chatbot to talk through a bad day, treat any "I feel anxious" or "I was traumatized" reply as generated text, not a real feeling, and do not let it set your own mood.
• For anything about your mental health, use a person or a service that is built for it, and keep the chatbot for practical help such as drafting a message or finding a number.

Source: The Decoder. This is our short summary of their reporting. The paragraphs marked "Our read" and the things to do are SeedFoundry's opinion. The summary is drafted with AI tools. The picture is our own graphic. How we source and credit