Missouri's restriction on marketing artificial intelligence as a mental health professional takes effect on 28 August 2026. Section 407.3007 of HCS SB 1019 bars a person or entity that develops or deploys AI in the state from advertising or representing that the system is, or can act as, a mental health professional. It also covers claims that the AI can provide therapy, psychotherapy, or a mental health diagnosis.
The measure is narrower than a general ban on AI in mental health. Its text targets public claims about professional status and capability. That distinction matters for product, legal, compliance, and marketing teams because the control point is not only model behavior. It includes the words used on websites, app listings, sales pages, onboarding screens, demonstrations, and other public materials.
What the enacted text requires
Missouri's official bill record identifies HCS SB 1019 as signed by the governor with an effective date of 28 August 2026. Governor Mike Kehoe's office separately confirms that SB 1019 was signed on 13 July.
The operative provision appears in section 407.3007 of the final House Committee Substitute. It defines AI broadly and then establishes three linked rules:
- Developers and deployers in Missouri must not advertise or represent AI as a mental health professional.
- They must not claim that AI can provide therapy, psychotherapy, or a mental health diagnosis.
- A violation is an unlawful practice under the Missouri Merchandising Practices Act.
The statute assigns enforcement to the attorney general. Individuals may report suspected violations. If the attorney general finds that a violation occurred, the text says the attorney general shall start a civil action. A court may award damages, civil penalties, injunctive relief, legal fees, and other appropriate relief. The specified civil penalties are $10,000 for a first violation and $20,000 for a subsequent violation.
Those figures and duties come from the enacted bill text. They should not be expanded into claims about how a court will interpret a particular website, chatbot exchange, or product design. This article is operational analysis, not legal advice.
What the law does not establish
The provision does not say that every AI feature used in a mental health setting is prohibited. It does not create an approval regime for all wellness tools, and it does not certify any product as safe when its marketing avoids the restricted language.
It also does not resolve every channel question. A public landing page is plainly advertising, but teams should obtain legal advice on less obvious cases such as generated in-product statements, user-created marketplace copy, third-party reseller claims, or content shown outside Missouri. The safe operational response is to identify those channels and assign them for review, not to assume that one approved homepage sentence covers the whole product.
Why this is a governance and evidence problem
Marketing restrictions are often handled as a copy review. AI products make that approach fragile because claims can appear in many places and change without a conventional campaign launch. A model-generated description may differ from the app-store listing. A partner may repeat an old capability statement. A sales demonstration may use language that the website no longer uses.
Teams therefore need a claims inventory linked to the AI-system inventory. Maetra's guide to mapping AI systems to compliance obligations explains how to connect a system, requirement, owner, control, and evidence. The AI compliance evidence checklist provides a practical structure for retaining the proof behind that decision.
For this Missouri requirement, the evidence record should connect each public claim to the product version, target market, content owner, approving reviewer, source of the capability statement, publication locations, and next review date. It should also record whether AI can generate or modify customer-facing descriptions without a human check.
A practical review plan
Teams offering conversational wellness, coaching, companion, or health features in Missouri can take five immediate steps:
- Identify scope. Record which legal entity develops or deploys the system, where it is offered, and which product surfaces reach Missouri.
- Inventory claims. Search current websites, app stores, advertisements, sales material, support pages, onboarding, and generated public text for therapist, therapy, psychotherapy, diagnosis, and equivalent claims.
- Verify capability language. Separate factual product functions from professional-status claims. Keep the source, reviewer, and approval date for every retained statement.
- Control changes. Require review before a model, marketer, partner, or product release introduces new mental health positioning.
- Preserve evidence. Retain the approved copy, captured publication locations, effective dates, removals, and monitoring results.
A keyword search alone is not enough. The law addresses representations, so a claim may create the same impression without using the exact statutory terms. Legal review should assess meaning in context.
Maetra analysis
Missouri's approach makes public positioning part of the control surface. A team can have careful model testing and still create exposure if its commercial language overstates professional status or clinical capability. Conversely, removing a prohibited phrase does not prove that the underlying product is safe, effective, or compliant with every other health, privacy, consumer-protection, or professional rule.
The durable control is a traceable claim process. Each statement should have an owner, a verified product basis, a jurisdiction decision, an approval record, and a review trigger when the model or experience changes. That structure helps teams respond to this law without converting a one-day copy edit into false confidence.