State AI disclosure laws in 2026: a plain-English guide for small businesses
By Antony Boguslavskiy · · 10 min read
The short answer
As of October 2026, a handful of state AI disclosure laws touch businesses that use AI with customers: California’s bot law, Utah’s generative AI disclosure rules and Maine’s chatbot law are the most direct. New Jersey has a bill but, as far as we found, no law yet. The practice that fits all of them: say plainly that it is AI, at the start, and offer a person.
As of October 2026. This is general information, not legal advice. Laws change, and how they apply depends on your business and where your customers are. Check with your attorney.
If an AI answers your phone or your website chat, you have probably wondered whether you are legally required to say so. The honest answer is: in some states, in some situations, yes. And the rules are spreading. This guide walks through the state AI disclosure laws that matter for a small business using AI on calls and chats, what each one actually says, and a simple practice that satisfies all of them.
We link the statute or the legislature's own page for every law, so you can read the source rather than our summary.
Why AI disclosure laws exist
AI disclosure laws exist because AI can now hold a conversation that sounds human, and lawmakers worry people will be misled about who, or what, they are dealing with. Most of these laws share one idea: a customer should not be tricked into thinking a machine is a person, especially when money, health or legal rights are involved.
They grow out of ordinary consumer protection. The Federal Trade Commission already polices "unfair or deceptive acts or practices in or affecting commerce" under the FTC Act, and pretending an AI is a person can be deceptive whether or not a state has an AI-specific law. The state laws below make the expectation explicit.
There is no single federal statute, as of October 2026 and as far as we found, that requires every business to disclose AI in customer conversations. Outbound calls are a separate matter: the FCC ruled in 2024 that AI-generated voices count as "artificial" voices under the Telephone Consumer Protection Act, so AI calls to consumers generally need prior express consent (FCC Declaratory Ruling, FCC 24-17). We cover that in TCPA rules for AI voice calls. This post is about disclosure in conversations, mostly on calls customers make to you and chats they start.
State AI disclosure laws in force, as of October 2026
Here are the state laws we found that require, or directly bear on, telling people they are dealing with AI. Read the "covers" column carefully: several are narrower than headlines suggest.
| State | Law | In force | Covers | What it asks |
|---|---|---|---|---|
| California | Business and Professions Code sections 17940 to 17943 (the bot disclosure law) | July 1, 2019 | Bots online that try to mislead people about being artificial in order to drive a sale or influence a vote | A clear, conspicuous disclosure that it is a bot |
| Utah | Utah Code Title 13, Chapter 75 (S.B. 226, 2025) | May 7, 2025 | Generative AI used with consumers in transactions, and in licensed professions | Disclose if asked; licensed professions disclose upfront in high-risk interactions |
| Maine | Public Law 2025, chapter 294 (LD 1727) | Enacted 2025 | AI chatbots used in trade or commerce | Tell consumers they are not dealing with a human, if the chatbot could mislead them |
| Colorado | S.B. 26-189, replacing S.B. 24-205 | Main duties from January 1, 2027 | Automated decision-making technology used in consequential decisions | Notice at the point of interaction, explanations after adverse decisions |
| New York | General Business Law Article 47 (AI companions) | November 5, 2025 | AI companion apps, not business customer service | Notice that the user is not talking to a human |
California: the bot disclosure law
California's law makes it unlawful to use a bot to communicate with a person in California online "with the intent to mislead the other person about its artificial identity" in order to incentivize a purchase or sale, or to influence a vote. A bot user avoids liability with a disclosure that is "clear, conspicuous, and reasonably designed to inform persons" that it is a bot. The law defines a bot as an automated online account and has been in effect since July 1, 2019 (Business and Professions Code 17940 to 17943).
Two things to notice. It is written for online interactions, so whether it reaches a phone call is not clear from the text. And it turns on intent to mislead. A website chat that introduces itself as a virtual assistant is doing what the law asks.
Utah: disclose when asked, and upfront for licensed professions
Utah has the most detailed rules for everyday businesses. Under S.B. 226 (2025), now Utah Code 13-75-103:
- Any business using generative AI to interact with a consumer in a consumer transaction must disclose that the person is dealing with generative AI and not a human if the person clearly and unambiguously asks.
- People in regulated occupations (licensed or state-certified by Utah's Department of Commerce) must prominently disclose AI use when it is a "high-risk" interaction. High-risk includes collecting health, financial or biometric data, or giving personalized advice someone could rely on for significant decisions, such as financial, legal, medical or mental health advice.
- That upfront disclosure must be spoken at the start of a voice interaction and written before a text one starts.
The law also offers a safe harbor: a business is not subject to enforcement under that section if its AI clearly and conspicuously discloses, at the outset and throughout the interaction, that it is generative AI, is not human, or is an AI assistant. Violations can bring administrative fines of up to $2,500 each from Utah's Division of Consumer Protection. The same bill extended Utah's broader Artificial Intelligence Policy Act to July 1, 2027.
Maine: chatbots in trade and commerce
Maine's 2025 law prohibits using an AI chatbot in trade or commerce "in a manner that may mislead or deceive a reasonable consumer into believing that the consumer is engaging with a human being" unless the consumer is told they are not engaging with a human. A violation counts as a violation of the Maine Unfair Trade Practices Act (legislature's enacted law summary, LD 1727).
Colorado: replaced and narrowed
Colorado's 2024 AI Act, S.B. 24-205, never took effect. In May 2026, the governor signed S.B. 26-189, which repeals and replaces it with rules for "automated decision-making technology" used in consequential decisions: education, employment, housing, financial services, insurance, health care and government benefits. Deployers must give "clear and conspicuous notice to consumers at the point of interaction" with a covered system, and explain its role after an adverse decision. Key obligations start January 1, 2027.
For a receptionist that answers questions and books appointments, this law is aimed at something different: automated decisions about people. If you use AI to screen tenants, job applicants or loan applicants, read it closely with your attorney.
New York: AI companions, not customer service
New York's General Business Law Article 47 requires operators of "AI companions" to tell users, at the start and at least every three hours in continuing interactions, that they are not communicating with a human. It took effect November 5, 2025. The definition excludes systems "used by a business entity solely for customer service," so a business answering calls about its services is generally outside it.
What they ask of a small business answering calls and chats
Put together, these laws ask a small business for three things: never let the AI claim to be human, answer truthfully when someone asks, and, in some situations, say it is AI before the conversation gets going. Whether the third applies depends on the state and what the conversation involves.
A plain-language way to check your own setup:
| Question | If yes |
|---|---|
| Could a reasonable caller think the AI is a person? | Disclose. That is the core of the Maine and California laws |
| Does the AI ever say it is a person, or dodge the question? | Stop. That fails every law above and risks FTC deception rules |
| Do you serve customers in Utah? | Make sure it answers honestly when asked. Licensed professionals: disclose at the start when the call involves health, financial or biometric data or advice |
| Is your business a licensed profession, such as health care, law or finance? | Disclose upfront, whatever state you are in. It is the safest reading |
| Does the AI make decisions about people, such as screening applicants? | Read Colorado's S.B. 26-189 and talk to your attorney |
| Do you record calls? | Check your state's recording notice rules too. That is a separate law from AI disclosure |
Remember that your callers' location can matter, not just yours. A New Jersey business with customers who call from other states may be subject to those states' rules. That is a question for your attorney.
New Jersey and New York: where things stand
As of October 2026, we found no New Jersey statute in force that specifically requires businesses to disclose AI in customer calls or chats. There is a bill. Assembly Bill A4730 would require businesses to notify consumers when they are communicating with generative AI in trade or commerce. It cleared the Assembly Science, Innovation and Technology Committee in March 2026, according to the Assembly Democrats' announcement. Bills change or stall, so check its current status on the New Jersey Legislature's website before relying on this.
New York's AI companion law applies to companion apps, not business customer service, as above. We found no New York statute in force that generally requires a business to disclose AI on customer service calls.
Even without a state law, New Jersey and New York businesses are covered by general consumer protection law, and federal law against deception. If you serve customers across the New Jersey and New York area, the practice below keeps you on the right side of all of it.
A disclosure practice that works everywhere
The simplest approach is to meet the strictest rule everywhere: say it is AI at the start, never claim to be human, answer honestly when asked, and give people a way to reach a person. That satisfies Utah's safe harbor, Maine's and California's laws, and the expectation behind every bill we have seen.
In practice:
- Disclose in the greeting. "Thanks for calling Oak Street Dental, this is the virtual assistant. How can I help?" Four words, no real cost.
- Never pretend. The AI should never say it is a person, use a human name as if it were staff, or dodge "Am I talking to a real person?"
- Offer a person. "I can help, or connect you with the team." A route to a human is good service, and Meta, for one, requires it for automated WhatsApp replies (WhatsApp Business Messaging Policy).
- Do the same in chat. The chat window should say it is a virtual assistant before the first reply.
- Write it down. Note what your AI says and when you set it up. If a question ever comes, you can show what callers heard.
- Review it yearly. Put a reminder in your calendar to check for new laws in the states you serve.
For how callers actually react to disclosure, and how to word it so it does not sound cold, read will customers know they're talking to AI?
How Vocenya's disclosure settings work
Vocenya's AI receptionist has two disclosure settings, on the Receptionist page under Language & AI disclosure. The default, and the one we recommend, is to say it is a virtual assistant upfront. The other option is to say so only when a caller asks. In both modes it tells the truth when asked and never claims to be a person.
| Setting | What callers hear | Fits |
|---|---|---|
| Say it's a virtual assistant upfront (default) | The greeting includes "this is the virtual assistant" | Every business, and the safe choice for licensed professions and for callers from states like Utah and Maine |
| Only say it's AI if the caller asks | A plain greeting; it confirms it is a virtual assistant when asked | Businesses comfortable that their callers and their states' rules allow it |
Website chat introduces itself as your virtual assistant and says a person from your team can take over. The steps of a call, including what the receptionist says about itself, are in what happens on a call. One more thing that article notes: Vocenya does not play a "this call may be recorded" message, so check what notice your state requires and add it to your greeting if needed.
If you are not sure which setting fits your business, choose upfront. You can see plans and the 14-day free trial on the pricing page.
Not legal advice. This summary reflects the sources linked above as of October 2026. Confirm current law with your attorney before relying on it.