More therapists are turning to artificial-intelligence tools that record sessions, generate transcripts and draft clinical notes, promising to cut paperwork and let clinicians spend more time with patients. Startups such as Berries pitch HIPAA-compliant platforms that process audio in real time, delete recordings and refrain from using session content to train models. Subscription fees typically range from $19 to $99 a month.
The adoption is testing the bounds of patient trust. One patient described feeling violated after realizing a session was recorded despite her hesitation, underscoring the centrality of consent and privacy in talk therapy. Surveys show Americans remain wary: a YouGov poll finds just a small minority open to AI for mental health care, while KFF reports broad concern over how health data are stored and used by AI systems.
Ethicists warn that HIPAA compliance doesn’t eliminate breach risks, and that consent forms alone may not constitute meaningful consent. Clinicians also worry that an “AI listener” changes the room’s dynamics and that errors in auto-generated notes could enter the medical record and pose legal exposure.
Proponents cite sizable productivity gains—mirroring results from ambient AI scribes in broader medicine—and say clear opt-in discussions and careful review of notes can mitigate risks. The technology’s future in mental health may hinge on transparent consent, robust safeguards and clinician oversight.
Related article:
– Ethics and governance of artificial intelligence for health
