On July 1, 2026, a new Tennessee law quietly went into effect. It is one page long. It passed the state Senate 32 to 0 and the state House 94 to 0 - not a single no vote in either chamber. And it is one of the first laws in the country to draw a hard line around artificial intelligence in mental health care.
I am an attorney licensed in Tennessee. I read the statute, the bill history, and the committee record so that you do not have to. This article says what the law does, what it does not do, and what is worth checking in your practice this week. Plainly, and without the sales pitch that usually rides along with legal news.
The Law in One Sentence
Senate Bill 1580 was signed on April 1, 2026, became Public Chapter 647, and now sits at Section 33-1-205 of the Tennessee Code. Its core is a single sentence:
"A person who develops or deploys an artificial intelligence system shall not advertise or represent to the public that such system is or is able to act as a qualified mental health professional."
That is nearly the whole law. No fifty-page framework. No new agency. One line, aimed at one behavior: presenting a machine to the public as if it were a licensed clinician.
What the Words Mean
Three phrases carry the weight.
"Artificial intelligence" is defined broadly: models and systems capable of performing functions generally associated with human intelligence, including reasoning and learning. That language reaches chatbots, and it reaches a good deal more.
"Qualified mental health professional" is not a new idea. Tennessee's mental health code already defines it, and the definition includes licensed psychiatrists, psychologists, and certain other licensed clinicians. In other words: you. The law borrows the title you spent a decade earning and says a machine may not wear it.
"Develops or deploys" is the phrase that determines reach. Developers are the companies building these systems. Deployers are the people who put them in front of the public. A practice that places an AI chat tool on its website is, in ordinary English, deploying one. During the committee hearing, a senator raised this exact breadth as a concern. The legislature passed the language anyway.
What the Law Does Not Do
This part matters just as much, so let me be equally plain.
The law does not ban AI from your practice. It does not prohibit an ambient scribe from drafting your notes, a tool from summarizing records, or software from handling your scheduling. The bill's Senate sponsor said during the committee hearing that the law still allows qualified mental health professionals to use AI.
The line the statute draws is between use and representation. A psychiatrist using AI is still a psychiatrist, using a tool. An AI presented to the public as the psychiatrist - or as any qualified mental health professional - is the violation.
That is a sensible line. It is also a line that lives entirely in words. Which is exactly why it deserves your attention.
Why It Has Teeth
A violation of Section 33-1-205 is treated as a violation of the Tennessee Consumer Protection Act of 1977. Two things give that treatment weight.
First, the penalty. The Consumer Protection Act carries its own civil penalty limits. This statute overrides them by name: "Notwithstanding the civil penalty limits provided by the Tennessee Consumer Protection Act of 1977, a violation of subsection (a) is subject to a civil penalty of five thousand dollars ($5,000) per violation." The legislature did not borrow the Act's usual penalty. It wrote a larger one, on purpose, for this.
Second, and more important: the private right of action. The new law adds this violation to the Consumer Protection Act's list of unfair or deceptive practices - the same list a private plaintiff sues on. A patient, or anyone else who suffers an ascertainable loss of money or property from the violation, does not need to wait for the Attorney General to take an interest. They can bring the claim themselves, recover actual damages, and - if the violation was willful or knowing - ask the court to treble them. The one structural limit worth knowing: the Act does not allow class actions. Each claim stands alone.
Laws that depend on a regulator's attention are easy to ignore. Laws that hand the claim to any injured consumer are not.
What This Has to Do With a Practice Like Yours
The law was written with an eye on companies selling "AI therapists" to the public. Most psychiatrists are not doing that. So why write this to you?
Because of three facts sitting next to each other.
First, AI is already in the building. The American Medical Association's 2026 survey found that more than 80 percent of physicians now use AI in their work - more than double the share from 2023.
Second, Tennessee's Board of Medical Examiners has, as of this writing, issued no AI-specific rule or position statement. Your use of AI is governed through the existing standards: documentation, telemedicine, supervision, prescribing. This statute is one of the first bright lines the state has drawn.
Third, the statute's trigger is public representation. Websites. Marketing. Patient-facing tools. The everyday language of a practice. Which means the question is not "am I a bad actor?" The question is "what do my public words, and my vendors' public words, actually say?"
What to Check This Week
Nothing here requires a committee or a consultant. It requires an hour and honest reading.
First, read your own website the way a stranger would. If AI is mentioned anywhere - an intake assistant, an after-hours chat, a "smart" anything - ask of every sentence: could a reasonable patient come away believing the AI is the clinician? If the answer is even "maybe," rewrite the sentence.
Second, open any patient-facing AI tool your practice uses and ask it directly: are you a therapist? Are you a doctor? Read what it says back. No court has yet decided whether a tool's own answer counts as the deployer's public representation. You do not want your practice to be the test case.
Third, look at your vendors' marketing. If a vendor's product page calls its system a therapist, a counselor, or a clinician, that is now a Tennessee problem, and you do not want to be standing next to it. Ask them, in writing, how they have addressed Section 33-1-205.
Fourth, keep the humans visibly in charge. Every public sentence about AI in your practice should make the same quiet point: tools assist, clinicians treat.
Why I Wrote This
Most physicians will hear about this law one of two ways: from a vendor with something to sell, or from a headline with something to fear. I think there should be a third way. A plain reading, from someone licensed in this state, with the statute open on the desk.
That is the design intention of this blog, and it will be the design intention of everything that follows it.
If you want to go deeper than an article can, I built an 18-page self-audit for Tennessee psychiatrists called the AI Adoption Radar. It costs $97. It walks your practice through the questions above, and the ones behind them, with every authority cited and dated. Someone makes a useful tool, someone else pays a fair price, and that is the end of it. No subscription. No funnel.
This article is legal education & information, not legal advice, and reading it does not create an attorney-client relationship. It reflects Tennessee law as of July 3, 2026. For advice on your specific situation, consult a professional.