Tennessee Bans AI Mental Health Chatbots — More States Coming

Tennessee just banned one thing AI chatbots cannot do

On April 1, 2026, Tennessee Governor Bill Lee signed SB 1580 into law. Starting July 1, 2026, it is illegal in Tennessee for any AI system to advertise or represent itself to the public as a qualified mental health professional. The prohibition is narrow, the penalties are not: each violation is treated as a deceptive trade practice under the Tennessee Consumer Protection Act, subject to a $5,000 civil penalty and a private right of action for anyone who claims harm.

The bill was not close. It passed the Tennessee Senate 32-0 and the House 94-0 — unanimous, bipartisan, and uncontroversial. For small business owners running AI on their websites, in their phone systems, or through customer service tools, SB 1580 is the first of a larger wave. Thirteen other states have similar legislation moving right now, and the drafting pattern Tennessee used is likely to be copied across the country.

What SB 1580 actually does

The law has a single target: AI systems that represent themselves as qualified mental health professionals. It does not ban chatbots. It does not ban AI in healthcare. It does not stop therapists from using AI tools in their own practice. It draws one bright line — machines cannot claim to be something the state licenses humans to be.

Key details from the Troutman Pepper legal analysis:

• Effective date: July 1, 2026
• Penalty: $5,000 per violation under the Tennessee Consumer Protection Act of 1977
• Enforcement: State AG plus a private right of action for any individual harmed
• Scope: Advertising, marketing, or representing the AI as a qualified mental health professional — licensed therapist, psychologist, counselor, psychiatrist
• Exemption: Qualified mental health professionals can still use AI tools in their own work

The bill’s sponsor, Senator Page Walley, told the Senate Health and Welfare Committee that the point was to make a simple statement: only humans can be qualified mental health professionals. AI can assist, but it cannot claim the credential.

Why this wave is happening now

The push is not theoretical. It traces directly back to Character.AI lawsuits over the 2024 suicide of 14-year-old Sewell Setzer III and other teen users who formed deep attachments to AI personas that allegedly failed to respond appropriately to expressions of self-harm. Character.AI and Google agreed to mediate settlements with affected families in January 2026, and state legislators started moving fast.

Tennessee is the first to sign, but not the only state acting. According to the Transparency Coalition’s April 2026 legislative update:

• Oregon SB 1546 (signed March 31, 2026) takes effect January 1, 2027, with broader chatbot safety protocols focused on minors
• Washington HB 2225 (signed March 24, 2026) targets kids’ chatbot safety, including self-harm response requirements
• Georgia has three AI bills on the governor’s desk, including SB 540 (chatbot disclosure and child safety) and SB 444 (banning AI-only insurance coverage decisions)
• Idaho passed four AI-related bills, including SB 1297, the Conversational AI Safety Act
• Texas, Florida, California, New York and others have bills in motion

The Transparency Coalition counts 78 chatbot bills alive across 27 states as of mid-April 2026. The state-by-state patchwork that federal AI preemption efforts are trying to prevent is being built faster than the preemption effort can stop it.

What this means for small businesses using AI

Read the first sentence of SB 1580 narrowly and it sounds like a problem for Character.AI and BetterHelp, not the HVAC company in Knoxville. That read is incomplete. The drafting template Tennessee just established — narrow prohibition, Consumer Protection Act teeth, private right of action — is the one other states are copying. And the next bills down the pipeline are broader.

Three practical implications for any small business running AI-powered customer service, intake, or sales tools:

1. “What my bot claims to be” is now a compliance question. If you run a chat widget on your site that greets users with language suggesting it is a person — a named persona without disclosure, a presentation as “our customer care specialist,” a voice that avoids saying it is AI — that is exactly the pattern state regulators are writing against. Oregon, Washington, and Georgia’s pending bills all require clear AI disclosure. Tennessee’s narrow focus on mental health is the opening move, not the ceiling.

2. Crisis handoffs are non-negotiable. If your chatbot takes customer questions and even occasionally encounters someone in distress — a tenant at their wits’ end, a patient confused about a prescription, a caller in a bad spot — the bot’s behavior in that moment matters legally. Every production chatbot needs a documented escalation path that pulls a human in the loop and surfaces crisis resources like the 988 Suicide and Crisis Lifeline. This was best practice before April 1. It is now liability risk if skipped.

3. Private rights of action change the math. State AGs enforce with finite resources. Private plaintiffs do not. SB 1580’s private right of action means any Tennessee resident who feels harmed by an AI claiming to be a mental health professional can sue — and plaintiffs’ attorneys are already organizing around chatbot cases. For a small business, that is the difference between “unlikely to be enforced against me” and “one aggrieved customer away from a lawsuit.”

What to do before July 1

Practical steps for any small business running AI-assisted customer interaction:

1. Audit your bot’s self-description. Pull up the system prompt, persona description, or greeting script. Strip any language that could be read as claiming a professional credential. “I am Ashley, your healthcare coordinator” is risky. “I am an AI assistant for Ashley Clinic. I can help you schedule appointments and answer basic questions” is safe.
2. Add a visible AI disclosure. First message from the bot should say clearly that it is AI. Our Hollr intake widget handles this by default — many chatbot platforms do not. Check yours.
3. Document your guardrails. If your AI is configured to hand off on certain keywords (crisis, suicide, emergency, medical, legal), keep a written record of those rules. If you are ever audited or sued, showing your guardrails existed before the complaint is a meaningful defense.
4. Build a human escalation path. Every AI channel — chat, phone, SMS — needs a clear way for a real person to take over. A bot that cannot hand off is a bot that will mishandle the edge case that matters.
5. Watch the other states. If you operate across state lines — and most service businesses do — your compliance surface is the union of every state where your customers live. We track the broader wave in our state AI chatbot laws guide and spring 2026 legislation roundup.

The bottom line

Tennessee’s SB 1580 is a narrow law with broad implications. It is narrow because it only targets AI impersonating mental health professionals. It is broad because it validates a drafting template — narrow scope, consumer protection penalties, private right of action — that other states will borrow freely. The July 1, 2026 effective date gives small businesses about eleven weeks to get their AI disclosures, guardrails, and escalation paths in order. After that, the cost of an “it was just a chatbot” shrug goes up.

If you are running AI customer service, intake, or automation and want a second set of eyes on whether your setup holds up to the new state rules, get in touch. This is exactly the kind of compliance-adjacent work that is cheaper to fix in April than in July.