AMA pushes AI chatbot rules
- The American Medical Association on April 22 urged Congress to impose federal safeguards on mental-health chatbots, including disclosure rules, crisis-response systems and limits on tools that mimic licensed clinicians. - In its letters to House and Senate artificial-intelligence caucus leaders, the AMA said chatbots should not diagnose or treat mental-health conditions without regulatory review and adverse-event monitoring. - The push lands as Connecticut’s 2026 session ends May 6 and Grow Therapy says clients have already sent more than 800,000 messages to its new AI coach. (ama-assn.org)
The American Medical Association is asking Congress to put new federal rules around mental-health chatbots as more patients use AI tools for support between or outside therapy sessions. (ama-assn.org) In letters dated April 22, the AMA told the co-chairs of the House and Senate artificial-intelligence caucuses and the Congressional Digital Health Caucus that current safeguards are too uneven. (ama-assn.org 1) (ama-assn.org 2) The group wants chatbots to clearly tell users they are interacting with AI, not a person, and to bar systems from presenting themselves as licensed clinicians. It also wants federal agencies to police deceptive marketing. (ama-assn.org) The AMA also said chatbots should not diagnose or treat mental-health conditions without regulatory review, and that Congress should clarify when an AI tool counts as a medical device. (ama-assn.org) (medicaleconomics.com) Another request is crisis protection: developers should build systems that detect self-harm risk, use de-escalation language and refer users to appropriate resources. The AMA also called for ongoing safety monitoring, adverse-event reporting and stricter standards for tools used by children and adolescents. (ama-assn.org) The letters frame AI chatbots as a gap-filler in a strained mental-health system, not a replacement for clinicians. The AMA said patients still face access and affordability problems, and that well-designed tools could help connect people to care if guardrails are in place. (ama-assn.org) The timing is notable because companies are already shipping products built around that in-between-sessions pitch. On April 24, Grow Therapy announced an in-app AI coach and said patients had already sent more than 800,000 messages through it. (prnewswire.com) Grow Therapy said its coach is monitored with safety features developed by licensed clinicians and is “not intended to diagnose conditions or provide treatment.” The company described it as an optional support tool for the other 167 hours of the week outside a typical one-hour therapy visit. (prnewswire.com) At the state level, Connecticut’s 2026 regular session runs through May 6, creating a near-term deadline for any bill lawmakers want to pass before adjournment. (cga.ct.gov 1) (cga.ct.gov 2) That leaves Congress, regulators and statehouses working on rules while companies keep testing clinician-supervised AI tools in the market. The central fight is no longer whether mental-health chatbots exist, but how much disclosure, review and crisis accountability they will face. (ama-assn.org) (prnewswire.com)