Table of Contents
Do not index
Do not index
If you're building or launching an AI version of yourself as a paid coaching subscription, you're entering largely uncharted legal territory. While there's no single federal law that says "AI coaches must disclose," a patchwork of state regulations and FTC rules creates real compliance requirements that could affect how you market and operate your AI twin.
These rules are moving fast. Colorado just implemented the nation's broadest AI disclosure law in June 2026, California added companion chatbot rules in January 2026, and Utah requires disclosure for certain high-risk AI interactions as of May 2025. Here's what these laws actually require and how they apply to AI coaching subscriptions.
This analysis covers U.S. state and federal law as understood in September 2026 and is not permanent legal advice.
Do You Legally Have to Tell Subscribers They're Talking to AI?
The answer depends on where your subscribers live and how your AI operates. Three states now have specific AI disclosure laws that could apply to coaching subscriptions.
Colorado Revised Statutes Section 6-1-1704 requires anyone doing business in Colorado to disclose when consumers interact with an AI system, unless that would be "obvious to a reasonable person." This became operative June 30, 2026, and applies to all Colorado residents who subscribe to your service. The law doesn't specify exactly how to disclose - no mandated wording or font size - but it must reach "each consumer who interacts" with the system. Burying disclosure in terms of service is risky.
California's SB 243 (effective January 1, 2026) targets "companion chatbots" specifically - AI systems with natural language interfaces, adaptive responses, anthropomorphic features, and the ability to sustain relationships across multiple interactions. An AI coach that remembers users, speaks as the expert, and builds ongoing relationships could potentially qualify, though whether routine educational coaching meets the undefined "social needs" test remains legally unresolved. If your AI qualifies and could mislead someone into thinking they're talking to a human, you must provide "clear and conspicuous notification" that it's artificially generated.
Importantly, SB 243 includes a private right of action: consumers harmed by violations can sue for injunctive relief, attorney's fees, and the greater of actual damages or $1,000 per violation. This creates real financial stakes beyond theoretical compliance concerns.
If your AI could qualify as a California "companion chatbot," there are additional requirements beyond basic identity disclosure. You must maintain and publish on your website a protocol to prevent the AI from producing suicide or self-harm content, with referrals to crisis services if users express ideation. For users you know are minors, you must also disclose AI status and, by default, repeat a break-reminder notification at least every 3 hours during continued use. Annual reporting on the suicide-prevention protocol begins July 1, 2027. This matters significantly for coaches whose niche touches emotional wellbeing, confidence, grief, relationships, or weight loss - not just overtly romantic or therapeutic applications.
It's worth noting that California has an older, narrower bot law: SB 1001 (Business & Professions Code Sections 17940-17943), operative since July 1, 2019. This earlier law only applies when a bot is used to communicate with intent to mislead someone about its artificial identity in order to knowingly deceive them into a commercial transaction or influence a vote - a much narrower, intent-based trigger than SB 243. No liability exists if you disclose bot status in a manner that's "clear, conspicuous, and reasonably designed to inform." These aren't the same law and are often conflated in discussions about California bot requirements.
Utah Code Sections 13-77-101 to 105 takes a different approach. For ordinary consumer transactions, you only need to disclose AI use if a user "clearly and unambiguously asks." But for "regulated occupations" or high-risk interactions involving personalized recommendations for significant decisions, you must disclose prominently that generative AI is used - verbally at the start of verbal interactions, in writing before written interactions begin.
The practical reality is that state laws can stack. A nationwide coaching subscription could simultaneously trigger Colorado's broad rule, California's companion-chatbot requirements, and Utah's disclosure standards depending on where subscribers live.
What Does the FTC Say About AI Coach Marketing Claims?
Even without specific AI disclosure laws, the Federal Trade Commission's general deception standards apply to how you market your AI coaching service. Under FTC Act Section 5, a practice is deceptive if it's likely to mislead reasonable consumers about something material to their purchase decision.
This means marketing that leads people to reasonably believe they're getting personal human access to you when they're not can be deceptive regardless of state AI laws. The FTC judges deception on the "net impression" of your entire presentation - words, images, design, and what you don't say.
The FTC Consumer Reviews and Testimonials Rule (16 C.F.R. Part 465) is a binding trade regulation rule that directly prohibits creating or disseminating testimonials that misrepresent whether the reviewer exists, actually used your product, or had the claimed experience. This bars AI-generated fake customer testimonials, composite stories presented as real individuals, or attributing AI-written praise to actual customers without their approval. The rule became operative October 21, 2024.
Additionally, you cannot offer customers compensation or any incentive that is expressly or implicitly conditioned on the review or testimonial being positive. You can ask for an honest review and offer a non-sentiment-conditioned incentive (such as "leave an honest review, positive or negative, get X"), but not "leave a 5-star review and get X."
The FTC Endorsement Guides (16 C.F.R. Part 255) are FTC interpretive guidance rather than independently enforceable regulations, but conduct inconsistent with them can still support a Section 5 deception case. These guides specify that testimonials citing specific results are generally understood to represent typical outcomes unless you clearly disclose what consumers generally achieve instead. Generic disclaimers like "results not typical" aren't sufficient on their own - you need to state what results are actually typical.
For endorsements, any material connection between an endorser and your business (payment, free access, affiliate commission, employment, personal relationship) that consumers wouldn't reasonably expect must be clearly and conspicuously disclosed. In interactive media, that disclosure must be hard to miss, not buried in a hidden profile link or footnote.
Can You Call Your AI Service "Coaching" Without Legal Issues?
The term "coaching" itself isn't legally regulated in most jurisdictions, unlike "therapy" or "counseling." However, several factors affect how you can market AI coaching services.
Colorado House Bill 26-1195 (effective August 12, 2026) specifically prohibits using AI in ways that imply outputs are provided by licensed psychotherapy professionals or that the AI provides actual psychotherapy. But it preserves exceptions for tools that clearly disclose they aren't substitutes for clinical care and offer services like coaching, self-help, or guided journaling.
If your coaching content touches emotional or mental wellbeing, adding a clear statement that your AI isn't therapy or a substitute for licensed care provides important protection. This is especially relevant for life coaches, relationship coaches, or business coaches who address stress, confidence, or personal development. The intersection with California's companion chatbot requirements becomes particularly important here, as AI coaching platforms focused on data privacy and security must navigate both identity disclosure and content safety protocols.
The bigger legal risk comes from capability claims rather than the "coaching" label itself. If you market your AI as "trained on my methodology" or "gives the same advice I would," you need evidence to support those claims before publishing them. The FTC requires advertisers to have a "reasonable basis" for objective claims. Understanding what actually matters beyond marketing claims about custom AI coaches helps separate substantive capabilities from promotional puffery.
Many experts are building AI coaching services as a strategic response to commoditization concerns, as explored in discussions about whether ChatGPT will replace business coaches. The key is ensuring your marketing accurately represents what your AI can and cannot do compared to direct human coaching.
Implementation Considerations for AI Coach Platforms
Building compliant AI coaching services requires attention to both technical capabilities and legal positioning. Platforms like BuddyPro let experts configure their AI's opening message, tone, and system instructions during setup, making it straightforward to build honest AI-identity disclosure directly into the product experience rather than relying on buried terms of service.
The AI can introduce itself consistently to every new subscriber with something like "I'm an AI trained on [Expert Name]'s methodology and content to provide coaching based on their approach." This approach addresses multiple compliance concerns simultaneously: it satisfies state disclosure requirements, prevents FTC deception issues, and sets appropriate expectations about the service's nature.
For coaches considering building their own AI platforms, understanding the hidden costs beyond initial development includes these ongoing legal compliance considerations. The regulatory burden of maintaining proper disclosures, safety protocols, and documentation across multiple state jurisdictions adds operational complexity that many experts underestimate.
What's the Practical Minimum for Compliance?
Given how fast these laws are changing, and how state requirements can overlap, a sensible baseline approach protects you regardless of exact legal triggers.
Never market your AI as providing "direct personal access" to you unless a human actually delivers that. Show AI disclosure before or during the first interaction, not buried in terms of service. Configure your AI to answer honestly if users directly ask whether they're talking to a real person or to you personally.
If your content addresses mental or emotional wellbeing, include a plain statement that the AI isn't therapy or a substitute for licensed care. For California compliance, if your AI could qualify as a "companion chatbot," establish and publish suicide-prevention protocols and consider break reminders for known minors.
Never publish testimonials or customer results you can't substantiate with real evidence, and never suggest showcased results are typical unless you can support that claim. Avoid incentivizing specifically positive reviews, though you can reward honest feedback regardless of sentiment.
Keep records backing any capability claims about your AI before you publish them. Claims like "trained on my content," "available 24/7," or "provides coaching based on my methodology" should be demonstrably accurate.
Clearly disclose any material connections with endorsers, including payment, free access, or affiliate relationships, in ways that are hard to miss in the medium where the endorsement appears.
The legal framework around AI coaching continues developing, with several questions remaining unresolved in case law. But building transparency into your AI's identity and capabilities from the start protects both your business and your subscribers while these rules keep changing.
Related Reading
Written by