AI Therapy Notes: Efficiency Tool or Breach of Trust?

A growing number of mental health therapists are turning to artificial intelligence tools to record therapy sessions, transcribe conversations, and generate clinical notes automatically. While the technology promises significant time savings for overburdened clinicians, it is raising profound questions about patient privacy, informed consent, and the foundational trust required for effective therapy, as NPR reports.

The Patient Experience

For Molly Quinn, a 31-year-old librarian from Fayetteville, Arkansas, the issue became personal during a November 2025 therapy session. When her therapist mentioned trying an AI note-taking tool, Quinn asked to research it first — she wanted to understand where her words would go. But halfway through the session, she noticed something different: her therapist wasn’t taking notes as usual, and an iPad was propped up. The session was being recorded.

“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,” Quinn told NPR. “I felt completely violated.”

She canceled her next appointment and eventually found a new therapist, setting a clear boundary from the start: no AI use. “The trust was gone,” she said.

How the Technology Works

Companies including Berries, SimplePractice, and Blueprint offer AI note-taking tools that record sessions in real time, transcribe them, and generate draft clinical notes for therapist review. Prices range from $19 to $99 per month. Tal Salman, Co-CEO of Berries, told NPR that audio is processed in real time and deleted immediately, transcripts are stored on HIPAA-compliant U.S. servers, and therapy content is not used to train AI models. “The clinician remains fully responsible for patient care and the final documentation,” Salman said.

Public Skepticism

Despite these assurances, public trust remains low. A YouGov survey from August 2025 found that only 11% of Americans are open to using AI for mental health, and just 8% trust it. Major concerns include lack of human understanding (53%), fear of inaccurate advice (50%), and data privacy (49%).

A separate KFF Tracking Poll from March 2026 found that 77% of Americans are concerned about the privacy of personal medical information provided to AI tools. Yet 32% of adults have used AI for health information, including 16% who sought mental health advice — a figure that rises to 28% among adults ages 18 to 29.

Expert Concerns

Kellie Owens, an assistant professor of medical ethics at NYU Grossman School of Medicine, warns that HIPAA compliance does not eliminate breach risk. “Regardless of what protections we have in place, that doesn’t mean data can’t be breached,” she told NPR. “There are plenty of systems that are fully HIPAA compliant that still experience major data breaches.”

Owens also emphasized that written consent forms alone are insufficient. “We have a wide body of research showing that a consent form on its own does not mean a person is making an informed choice,” she said. “Any time you are recording a conversation, that should require a verbal conversation that a recording is taking place.”

Marisa Cohen, a couples and sex therapist in New York, noted that the mere presence of AI alters the therapeutic dynamic. “Even the presence of AI changes the therapeutic experience,” Cohen said. “Clients know or feel like something else is listening to them. That awareness can subtly alter their disclosure.” She also raised concerns about accuracy, noting that if AI hallucinates content and a clinician doesn’t catch it, errors become part of a client’s permanent medical record.

The Efficiency Argument

For some clinicians, however, the technology has been transformative. Kym Tolson, a Virginia-based therapist who practices remotely, said AI has dramatically reduced her documentation burden. “Most clinicians spend about 10 hours a week on administrative tasks, and five to seven of that is documentation,” she told NPR. “With the AI system, I spend about two minutes per client where it used to take me 15 to 20.”

Tolson said the tool has “given me my life back,” though she acknowledged that AI can hallucinate and requires careful review. “The clinician has to be very careful. You have to double- and triple-check.”

Her experience aligns with broader healthcare data. An analysis of AI scribes at The Permanente Medical Group found that 7,260 physicians saved nearly 16,000 hours of documentation time over 15 months across 2.5 million patient encounters.

Professional Guidance

In response to the growing use of AI, both the American Psychological Association and the American Counseling Association have released ethical guidance emphasizing transparency, informed consent, data privacy, human oversight, and clinician accountability, as Blueprint.ai summarizes. Both organizations stress that clinicians remain fully responsible for all clinical decisions and documentation, even when AI tools are used.

What’s Next

The debate over AI in therapy highlights a fundamental tension between efficiency and confidentiality in healthcare. As Quinn put it: “We’re going to see breaches. 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.”

With professional organizations still developing enforceable standards and state licensing boards beginning to weigh in, the coming years will likely see increased regulatory scrutiny. For patients, the question is becoming a standard part of intake: Is AI in the room, and do I have a choice?