In recent years, ChatGPT has taken on a social role in society and become an integral part of people's everyday lives. The launch of the service in 2022 came in the aftermath of a global pandemic, and today the service is something people turn to for everything from information to therapeutic conversations. The purpose of this study is to investigate how trust can be characterised in relation to ChatGPT as a therapist, how it affects our interpersonal relationships and our information behaviours. The empirical data is collected through digital observation of a social forum where users share their therapeutic use of ChatGPT. The results show a high level of trust in the service and the capabilities of AI. ChatGPT is perceived as a good therapist with endless patience and good listening skills, which is often contrasted with human therapists. At the same time, there are still concerns about misinformation, ingratiation, and privacy issues. The analysis is based on one of Wilson's models of information-seeking behaviour and Habermas' theory of communicative action. Based on this, it is concluded that ChatGPT's ability to provide detailed answers at any time, at a low cost and with little effort on the part of the user, strengthens people's trust in AI and makes it the obvious choice as a source of information. ChatGPT is not only praised as something other than a human being with baggage and bias, but is also attributed with a strong humanity, which is perceived as necessary to create a strong relationship with the service. The study concludes with a discussion of the scientific, societal and ethical aspects of the conclusions. This discussion emphasises that research should take into account the relationship that many people have with ChatGPT today, and the trust that comes with it, and that companies need to be more transparent in their work and understand the breadth of how their service is used.