In November 2025, the American Psychological Association expressed deep concerns regarding the reliance of some individuals on GenAI chatbots and wellness apps to deliver psychotherapy or psychological treatment. The APA’s health advisory emerged in the wake of a number of reported cases of “AI psychosis” and suicidality after people interacted with AI chatbots when seeking psychological support, particularly during times of crisis.

As the ISTP call for papers makes clear, these are indeed dark times. And as social scientists who are now training to be therapists (via social work and psychoanalysis, respectively), we feel more politically committed than ever to theorizing what it is that AI therapy is missing—and/or the ways in which it is, in some cases, actively deleterious to patient wellbeing. However, we are also interested in considering: is there any way that technologies of care can ever be useful? If so, how and under what circumstances—practically, theoretically, politically, and/or clinically?

By theorizing patient interactions with AI therapy bots alongside the figure of the sexual robotic companion or sexbot, we consider how and why these technologies come to stand in for both panacea and downfall. What kinds of interactions do they actually offer? How do they fall short?

We argue that by thinking specifically about the role of the other in constituting the self, it is possible to understand both the possibilities and limits of these technologies of care. In this unravelling, we utilize Glissant’s notion of the importance of opacity, Bion’s discussion of negative capability, Scarfone’s theorization (through Simondon) of autopoeisis, and broader existentialist theorizing on the possibility of self-knowledge within specific environments—or self-production within the particular environments constituted by unique Others.