If you’re in therapy, there’s a new risk sitting in the room with you, and it isn’t another person. It’s an iPad quietly recording every word you say. That’s what happened to 31-year-old Molly Quinn, whose story was reported by Futurism AI, drawing on an NPR investigation into how AI scribe tools are quietly entering the therapy room.
Quinn realized halfway through a session that her trusted therapist wasn’t taking notes by hand like usual. “She wasn’t taking notes like she usually did,” Quinn recalled. “The iPad was just propped up.” An AI model was capturing the conversation. The questions hit her fast: Where were her words going? Who was processing them? Could they one day become training data?
“The more I thought about it, the more I just started getting more and more sick to my stomach,” she told NPR. “This person who I’m supposed to be able to trust with some very private and very intense emotions had just completely disregarded something I said I was not comfortable with. I felt completely violated.” The therapist offered to stop. Quinn left anyway. “The trust was gone.”
What’s actually happening
This isn’t a one-off. According to Futurism AI, therapists across the country are adopting AI tools for notetaking and transcript generation, mirroring a wave already underway among doctors. The pitch from AI vendors is familiar: let the software handle the paperwork so clinicians can focus on patients. It’s the same refrain the AI industry uses everywhere. Let us do the tedious stuff for you.
The problem is that therapy isn’t like other paperwork. It runs on trust and total privacy. Introduce a recording device and the entire dynamic shifts.
Why this matters
What stands out here is how badly the technology is outpacing patient comfort. A YouGov survey cited by NPR found:
- Only 11 percent of Americans said they’d be open to AI in mental health care
- Just 8 percent said they’d trust AI being used this way
- 40 percent said they don’t trust the technology at all
Those numbers tell you the industry is pushing a product most people haven’t agreed to. And the clinical risk is real. AI transcription tools still hallucinate, and made-up details have already started creeping into medical notes. In a therapy record, a fabricated line about a patient’s mood or history isn’t an inconvenience. It’s a potential misdiagnosis.
There’s a subtler cost too. “Even the presence of AI changes the therapeutic experience,” Marisa Cohen, a couples and sex therapist in New York, told NPR. “Clients know or feel like something else is listening to them. That awareness can subtly alter their disclosure.” She called the technology “essentially a third party” in the room.
The breach question
Vendors say they’ve locked things down. Tal Salman, CEO of a popular AI scribe tool for therapists called Berries, insists recordings are deleted immediately and transcripts sit on HIPAA-compliant US servers. Maybe that’s true. But “trust us” is a hard sell when the data in question is your most private confessions.
Quinn’s fear is the one worth taking seriously: a future data breach. “We’re going to see breaches,” she told NPR. “Maybe not tomorrow, maybe not next week. But in a few years? I think we’re going to see them. And I don’t want my therapy session to be part of that.”
What to do about it
If you’re in therapy, you don’t have to accept this silently. A few defensive moves:
- Ask directly whether AI tools are used in your sessions, before they’re switched on.
- Request that recording be turned off. You have the right, and a good therapist will honor it.
- Ask where transcripts are stored, how long they’re kept, and whether anything is ever used to train models.
- Get the answers in writing if you can.
Consent should come before the device is propped up, not after you notice it. The AI industry wants a seat in the therapy room, but as this case shows, it hasn’t earned the trust that seat requires. Until it does, the burden falls on patients to ask the hard questions first. You can read the full account at Futurism AI.