Mateusz, Dr. Paweł Szczęsny, and Dr. Katarzyna Mika-Łabuz published peer-reviewed article titled “Large language models (LLMs) as psychotherapists: an analysis based on psychodynamic psychotherapy theory” in Ethics and Information Technology (Vol. 28). And it has a huge cybersecurity element!

Emotional support is one of the most common uses of generative AI. The authors test whether chatbots can meet the conditions of psychodynamic therapy, such as the therapeutic relationship, transference and the therapist’s own reflection. They point to the ethical risks of treating LLMs as quasi-therapists. If you care for your data, you better not tell chatbot your deepest secrets…

Mateusz brought the topic to a wider audience in an interview for the leading Polish weekly magazines Tygodnik Powszechny (“Czatbot cię wysłucha. Ale nie będzie twoim terapeutą”, No. 09/2026) and he wrote a follow-up in a Spider’s Web essay (“Kochanie, porozmawiajmy o moich algorytmach”). Links are provided below:

Peer-reviewed article: https://link.springer.com/article/10.1007/s10676-026-09917-x
Interview in Tygodnik Powszechny: https://www.tygodnikpowszechny.pl/czatbot-cie-wyslucha-ale-nie-bedzie-twoim-terapeuta
OpEd in Spider’s Web: https://spidersweb.pl/plus/2026/06/czy-chatbot-zastapi-nam-psychoterapeutow