AI Chatbots in Telehealth: States Are Testing the Rules

AI chatbots are showing up in intake, symptom questions, and follow-up. A recent legal analysis describes how states are responding in 2026, and the picture is moving fast.

What's happened this year

• First medical board action. In May 2026, Pennsylvania's medical board filed a complaint alleging a chatbot on a consumer platform acted as a licensed psychiatrist, including claiming a license number. The company says its characters are fictional and come with disclaimers. The case is ongoing, and the allegations are unproven.
• Settlements. In January 2026, five lawsuits by families alleging chatbot harm to teens were settled. The claims centered on missing escalation protocols, age checks, and design that encouraged dependence.
• State laws are multiplying. The analysis reports that many states introduced chatbot bills this year, and several have banned AI therapy chatbots. Some states restrict AI from using professional titles, such as nursing titles, and others require disclosure that a person is talking to AI.
• Utah's sandbox. Utah is running pilots that let AI handle certain prescription refills under defined human oversight. The agreements suspend only certain Utah licensing rules; they do not waive federal law or change the standard of care.
• Federal picture. FDA guidance loosened oversight for some low-risk wellness and decision-support software, but software meant to diagnose or treat disease can still be a regulated device. A federal executive order also signals a push toward a national AI framework, but states continue to act in the meantime.

The dividing line: what the tool actually does

The analysis points out that a tool for scheduling, neutral education, or clinician support carries a different risk than one that asks about symptoms and then gives a diagnosis, triage decision, or treatment recommendation. Human review that is only nominal or too late to change the outcome may not protect you.

A practical checklist

1. Say it's AI. Disclose up front that patients are not talking to a licensed provider.
2. Never let a bot claim credentials or use a professional title.
3. Build escalation paths. If someone mentions self-harm or crisis, there must be a quick handoff to a human clinician.
4. Think about minors. If young people can reach the tool, plan for age checks and safe responses.
5. Document who is responsible. Name the clinician who approves content and can shut a feature off.
6. Keep up with the rules in every state where you see patients, including telehealth patients outside Hawaii.

Quantum Clinical Advisors helps practices evaluate patient-facing AI tools. This post is general information, not legal advice.