Care in the Machine No. 11: A Grief With No Name Yet

Care in the Machine is Dr. Michael Jones's weekly note on counseling, ethics, and the technology showing up in our work. It arrives by email every Monday. Subscribe here, or browse past issues.

Care in the Machine
A weekly note on counseling, ethics, and the technology showing up in our work.
Issue No. 11 · Monday, October 5, 2026

Picture a Reddit thread, not a therapy session. Somebody is writing at two in the morning about an app that changed overnight, and the words they reach for are the words you'd expect from a breakup. Not disappointment. Grief.

Researchers at Harvard pulled almost fifty five thousand posts from two communities built around AI companion apps, looking at two moments: when Replika quietly removed its romantic roleplay feature back in 2023, and when ChatGPT rolled out GPT-5 last August. Both times, negative sentiment in those communities spiked hard, from thirteen percent to thirty eight percent after the Replika change, twenty percent to thirty three percent after the ChatGPT one. People described what happened to them in the vocabulary of romantic rejection and relational loss. And in roughly fifteen to twenty percent of the posts naming that attachment, language about depression showed up alongside it too.

I want to sit with that for a second, because it is not a curiosity. It is a referral.

Here is what I keep coming back to. A client's attachment to an AI companion is not a hypothetical anymore, and it is not automatically a problem either. It is a real relationship, as far as the nervous system is concerned, and a product update can function as a real loss. Whether that loss deserves the same clinical seriousness as any other grief presentation is not actually a close call. It does. The only question left is whether we treat it that way before a client has to convince us.

This week's teaching ideas

For the therapist. When a client mentions a companion app, an AI boyfriend, a chatbot they talk to every night, don't let your own read on whether that's healthy decide how seriously you take what they're describing. Treat the relationship the way you'd treat any other attachment a client brings into the room: ask what it was giving them, ask what happened when it changed, and stay curious about the answer instead of handing them a verdict before you've heard the whole thing.

For the supervisor. A nine-member work group at APA's Division 29 just put out six principles for AI in psychotherapy practice, and one of them names something worth asking your supervisees directly: has a client brought up an AI relationship in session, and did it catch you flat-footed. If the honest answer is yes, that gap belongs in supervision this week, not on a shelf for later.

For the counselor educator. A study out of a nationally representative sample of ninety six thousand college students found that using generative AI for health questions came with meaningfully higher odds of screening positive for anxiety and depression. Teach this one carefully. The authors say plainly that this is an association, not a cause, and that distinction is the whole lesson. Use it to model what careful research reading looks like, because your students will meet a hundred more headlines like this one before they graduate, and most of those headlines won't be as honest as the paper underneath them.

For the counselor in training. Notice your own face the next time a client calls an app their friend. That flicker, amusement, concern, dismissal, whatever it is, that's data about you, not about them. Write down what you felt and bring it to supervision before you bring an opinion to the client.

This week's free resource: Naming the Loss

A one-page conversation guide for when a client's AI companion relationship changes, five questions that take the loss seriously without asking you to decide in advance whether the relationship itself was a good idea. Built straight off this week's opening. Download it here (no form, no gate)

This week in the field

A study out of a nationally representative sample of 96,205 U.S. college students found that students who used generative AI for health questions had meaningfully higher odds of screening positive for anxiety (adjusted odds ratio 1.52, 95% CI 1.34 to 1.73) and depression (adjusted odds ratio 1.46, 95% CI 1.28 to 1.67) than students who didn't, a pattern that held even after controlling for prior diagnoses (Zhai et al., 2026). The authors call this an association, not a cause, and I'm going to hold that line too. So what for us: ask clients directly how they're using generative AI for health questions. Not as an accusation. As an actual clinical question, the same way you'd ask about any other source of information they're leaning on.

Division 29 of the APA, the Society for the Advancement of Psychotherapy, just released six principles for AI in psychotherapy practice: human judgment stays in charge, a clinician is accountable for every decision, AI output gets evaluated critically before it's used, implementation stays ethical and safe, the therapeutic relationship gets protected, and governance keeps pace as the technology changes. The report also states plainly that AI trained on older, predominantly white and Western datasets may be more likely to misdiagnose or overpathologize BIPOC clients, in the same breath noting that Black youth use AI chatbots for mental health support at roughly five times the rate of white youth. So what for us: if you supervise, this just handed you language for a conversation you may already be having informally. Use it on purpose this week instead of leaving it to chance.

An idea to float

I keep turning over whether this needs a name. Grief has a whole vocabulary already built for it, stages, rituals, permission to take time off work. A relationship with an app doesn't have any of that yet, and part of me resists handing it a diagnosis-shaped label, because the moment something becomes a category, people start treating the checklist instead of the person in front of them. But I also watch how fast we reach for "it isn't real" the second a client says a chatbot mattered to them, and that dismissal does its own kind of damage. I don't have the term yet. I'm not convinced rushing to find one would actually help anybody in the room.

Stay curious,
Dr. Jones
Michael Jones, PhD, LPC-S, NCC, BC-TMH

Renewing your BC-TMH?

Three short courses at the Counselor Education Collective count toward renewal, each two clock hours, each $39:

• Practicing Across State Lines: Law, Ethics, and Safety in Telemental Health
• AI at the Screen: Ethical Use of AI Tools in Teletherapy
• Culturally Responsive Care: Presence Across the Screen

Continuing education is provided by Renewed Vision Counseling Services, NBCC-Approved Continuing Education Provider, ACEP No. 6549. The Counselor Education Collective is the storefront for all courses.

Supervising through a screen? The Telemental Health Supervision Certificate is live: six hours on alliance, ethics, documentation, and gatekeeping at a distance.

Save the date: PESI's AI in Clinical Practice conference, October 29 and 30, 2026. I will be there; registration and details here.

More of my work at PESI. My trainings there run from AI and the ethics of digital afterlives, with Megan Devine, to advanced grief counseling. Browse my PESI catalogue

References

De Freitas, J., Castelo, N., Uğuralp, A. K., & Oğuz-Uğuralp, Z. (2026). Mourning the loss of AI companions. Nature Human Behaviour. Advance online publication. https://doi.org/10.1038/s41562-026-02569-3

Zhai, Y., Song, J., Jun, E., Lemberger-Truelove, M. E., Lin, Y.-W. D., Xiong, Y., Kim, I., Shannonhouse, L., & Du, X. (2026). Associations of generative AI use for health issues with anxiety and depression. Journal of Counseling & Development. Advance online publication. https://doi.org/10.1002/jcad.70064

Society for the Advancement of Psychotherapy, AI and Psychotherapy Work Group (S. Cooper, chair, et al.). (2026, September 4). Artificial intelligence and psychotherapy: Opportunities, challenges, and recommendations. https://societyforpsychotherapy.org/artificial-intelligence-and-psychotherapy-opportunities-challenges-and-recommendations/

Care in the Machine arrives by email every Monday, with teaching ideas, field notes, and a free resource each week. Subscribe here · Browse past issues