Mental health therapists increasingly use AI for session notes and admin tasks
Growing numbers of mental health therapists are using AI for session notes and admin work, raising sharp divisions over privacy risks and therapeutic trust versus efficiency gains.
Objective Facts
A growing number of mental health therapists are using AI tools to record sessions, take notes and do administrative tasks. Companies offering similar services include SimplePractice and Blueprint, charging fees of $19 to $99 per month and appealing to solo practitioners overwhelmed by administrative burdens. Therapists using these tools report dramatic time savings—from 15-20 minutes per client down to about two minutes when manually handling the 10 hours per week typically spent on administrative work and documentation. However, patient and clinician concerns run deep. A YouGov survey found that only 11% of Americans would be open to AI for mental health care and just 8% trust it. Medical ethics expert Kellie Owens at NYU notes that HIPAA compliance does not eliminate breach risk, with many fully compliant systems experiencing major data breaches. Owens warns that if patients feel privacy is compromised, it can cause real damage to the therapeutic relationship.
Left-Leaning Perspective
Privacy advocates and patient safety experts raised the core concern about therapist AI adoption in recent coverage. Psychology.org highlighted personal accounts from therapists like one who, after reviewing privacy policies for AI documentation tools, found "vague language around data storage and third-party sharing" and decided against adoption. Marisa Cohen, a couples and sex therapist in New York, told NPR that introducing AI into therapy fundamentally changes the experience by introducing an electronic third party, raising "additional questions about trust and safety." Kellie Owens, an assistant professor of medical ethics at NYU Grossman School of Medicine, emphasized to NPR that HIPAA compliance "does not eliminate the possibility of breaches," noting that many fully compliant systems still experience major data breaches. Owens warned that "if patients feel that privacy has been compromised, that can do real damage to the therapeutic relationship." Left-leaning concerns also focus on clinical accuracy failures. Kevin MD published analysis noting that AI-generated notes have "inserted details that never occurred in the session"—including reports of suicidal ideation, substance abuse, or past trauma patients never mentioned, and in one case falsely documenting childhood sexual abuse. This mirrors warnings from Quill Therapy Notes, which emphasized that while AI offers efficiency, full session recording stored on third-party servers introduces "significant privacy concerns," with data breaches remaining "a persistent threat" despite security assurances. The left's position emphasizes that efficiency gains cannot justify risks to the foundational element of therapy: trust. Therapists on this side acknowledge efficiency benefits but argue they pale against potential harms to the therapeutic relationship and patient privacy. The framing prioritizes patient safety and informed consent over administrative convenience.
Right-Leaning Perspective
Therapist-advocates and efficiency-focused commentators emphasized liberation from administrative burden as the primary benefit. Kym Tolson, "The Traveling Therapist," told NPR that the AI system "has given me my life back," describing how time spent on documentation dropped from 15-20 minutes per client to about two minutes. She noted the flexibility has allowed her to maintain work-life balance and spend more time focused on clients rather than paperwork. APA Services reported that about 1 in 10 psychologists already use AI monthly for administrative work, and Damien Adler, a psychologist and co-founder of Zanda Health practice management software, told APA Services that "No one gets into mental health care to do paperwork, but administration is a massive part of the job and something that can drain a lot of energy," and "AI can now significantly reduce that burden." Right-leaning coverage also emphasized emerging research support. The Upheal research summary cited a 2022 cross-study of 872 individuals showing the public generally has positive views of AI in psychotherapy when supervised by humans, and noted that therapists recognize AI as reducing workload and potentially lowering human errors in documentation. Stanford HAI researchers, quoted in multiple sources, framed AI documentation as a "supercharged secretary" that can help clinicians with basic tasks, and proposed a staged approach to integration that prioritizes proper evaluation. The right's position frames the concern as one of responsible implementation rather than fundamental opposition. Proponents argue efficiency gains directly benefit patients by allowing therapists more presence in sessions, and that well-vetted tools with proper oversight solve privacy concerns.
Deep Dive
The therapist AI documentation story reflects a genuine tension between two legitimate healthcare values: administrative efficiency and patient privacy/therapeutic integrity. Both critiques and endorsements come from credible sources within healthcare. The NPR reporting published May 26, 2026, shows this is not an abstract debate—it's happening now in practice, with real companies (Berries, SimplePractice, Blueprint) selling these tools to real therapists at scale. What each side gets right: Privacy advocates correctly identify real vulnerabilities. AI hallucinations in clinical notes are documented problems, and therapy notes containing false information about suicidal ideation or trauma can have serious legal and clinical consequences. The concern that HIPAA compliance alone is insufficient is technically sound—breaches happen to fully compliant systems. And the point about therapeutic safety is philosophically coherent: therapy does depend on perceived privacy, and adding electronic recording changes that perception even if technically secure. What efficiency advocates get right: Therapist burnout from documentation is real and well-documented, with therapists spending 5-7 hours weekly on notes alone. This exhaustion can degrade the very clinical presence that makes therapy effective. The evidence that properly supervised AI tools reduce clinician burden while maintaining accuracy (when properly reviewed) is also sound. And the research showing patients appreciate reduced therapist stress is credible. What each side downplays: Critics tend to understate the genuine clinical cost of documentation burden—burned-out therapists are less present, and that harms care. Advocates tend to minimize that patient consent, however transparently obtained, cannot fully address power imbalances or the reality that once data is recorded, breach risk persists indefinitely. Neither side adequately addresses the missing middle ground: therapists could use AI for summarizing manual notes (as some vendors offer) rather than recording sessions, which would preserve some efficiency gains while reducing the persistent data vulnerability of continuous recording. What to watch: The regulatory response will be critical. As of May 2026, professional guidance from APA, NASW, and state licensing boards is still emerging. Whether regulators impose standards on vendors (security, audit trails, training requirements) or allow market fragmentation will determine whether this becomes a best-practice tool or a high-risk workaround. Also critical: whether actual breaches occur. If therapy session AI recordings are breached, the narrative will shift decisively toward the critics.