If you’ve got an AI agent answering your phone or replying to text messages for your business (or you’re thinking about setting one up), there’s something you should know.

A handful of states now have laws that say if a customer is talking to an AI, they have to be told they’re talking to an AI.

One of them has fines of up to $20,000 per violation (although, as you’ll see below, that one probably doesn’t reach a business that’s just answering its own phone). And more states are passing these laws every year.

The good news? For most businesses, the fix is one sentence. It’s about 10 words long, and your AI agent can say it in roughly four seconds.

In this post, I’ll walk you through what these laws actually say, which states have them, and exactly what to put in your AI agent so you’re covered pretty much everywhere.

A quick disclaimer before we start: I’m not an attorney, and nothing in this post is legal advice. I’m just explaining what I found while digging through this stuff for my own business. The laws on this are changing fast, so some of what you read here could be out of date by the time you see it. If you’ve got a real legal question about your situation, run it by an attorney. Near the end, I’ll show you a couple of easy ways to check what’s current.

What Is an AI Receptionist, Anyway?

At the time of this writing, there are basically two kinds of AI agents that small businesses are using right now.

1. Voice AI agents. This is software that answers your phone and talks to the caller in a pretty normal-sounding voice. It asks questions, listens to the answers, and (if you’re using the voice AI agent built into Stride CRM) plugs that information right into your CRM, which is just the database where you keep track of your leads and deals.
2. Chat AI agents (sometimes called conversation AI agents). These do the same job, but in writing. They reply to text messages, run the chat box on your website, or answer people on Facebook Messenger, WhatsApp, or wherever they’re messaging you.

Why So Many Real Estate Investors Are Paying Attention to This

If you’ve ever sent out direct mail or any other kind of outbound marketing, you already know the problem.

When you send out a few thousand letters, texts, or calls, the responses don’t show up on your schedule. They come in during dinner, while you’re at your day job, at 9:30 on a Saturday night, or at 3:00 a.m. on Christmas morning. Every one of those calls that goes to voicemail is potentially a seller who shrugs, moves on with their day, and never thinks about you again.

For years, the answer was to hire a call answering service or a virtual assistant. Basically, you’re renting a human being (usually someone you’ve never met) to pick up your phone and read a script you wrote for them. They’re not a sales rep. They don’t know much about your business. They can’t really help beyond what’s on the script.

RELATED: PATLive Review: How Helpful Is a Virtual Receptionist for Busy Entrepreneurs?

I’m not knocking the people who do that work. It’s still a viable option. But it costs more, and the quality depends a LOT on who picks up the phone that day. Did they wake up on the right side of the bed? Are they in a good mood? Did they remember to ask the questions you really need answered?

An AI agent answers on the first ring, at any hour. It asks the same questions in the same order every single time. And it never gets short with anybody because it’s been a long day.

Is an AI Agent as Good as You?

No, it’s not, and I’m not going to pretend it is.

But that’s not the right comparison. I’m not comparing my AI agent to you or me picking up the phone on our best day. I’m comparing it to a rented human reading my script. When you measure it that way, I don’t think the AI agent is that far off, and in some cases, I think it’s actually better.

If you want a real example and want to decide for yourself, give mine a call: (616) 504-6909. Throw whatever you want at it and see how it does.

The Biggest Objections (and the One That’s Legit)

A lot of people aren’t ready to try the AI agent thing yet. The objections I hear most often are:

• “Sellers are going to hate it.”
• “It feels impersonal.”
• “It’s kind of deceptive to have a machine talking to people like it’s a person.”

I think the first two have some validity, partly because people are comfortable with the way they’ve always done things. The thing you have to compare an AI agent to is what’s actually happening to those calls today, and for most small businesses, that’s voicemail.

That last objection, though? I think it’s totally legit. And it’s exactly where the law is starting to show up.

Which States Have AI Disclosure Laws?

A while back, somebody forwarded me a newsletter from an attorney about some new AI rules in Colorado. I’ll be completely honest with you, the way it was written made it sound like the sky was falling.

So I spent a bunch of time reading through what all the states are doing. What I found is that it’s not nearly as scary as some attorneys make it sound… but it is real, and it is growing.

The basic idea behind all these laws is the same: a person shouldn’t talk to an AI without knowing they’re talking to an AI.

As I write this, four states have laws that can reach a regular small business like yours or mine, and a fifth is on the way.

• Maine is the broadest. It covers phone calls and text messages, and it says you can’t use an AI chatbot in a way that might make a reasonable person think they’re dealing with a human unless you tell them. In Maine, a customer can actually sue you over it.
• New Jersey covers online bots used to sell merchandise or real estate. The bot must disclose what it is at the start of the conversation. Fines start at $2,500 and go up to $10,000 for repeat offenses.
• Utah says if somebody clearly asks whether they’re talking to an AI, it has to tell the truth. That’s $2,500 per violation. But Utah also says that if your AI identifies itself at the start and throughout the conversation, you’re protected.
• California makes it illegal to use a bot online to trick somebody into buying something. If the bot says what it is, you’re in the clear.
• Colorado has a law that kicks in on January 1, 2027, and it’s where those $20,000 fines come from (that’s the max penalty per violation under Colorado’s consumer protection law, and each consumer counts as a separate violation). But this law got a LOT narrower in 2026. Colorado replaced its original AI law with a new one that’s aimed at AI used to make “consequential decisions” about people, like whether they get a loan, a job, housing, insurance, or health care. If your AI agent just answers the phone and takes down a seller’s information, it most likely doesn’t apply to you. The attorney general is still writing the final rules, so it’s worth keeping an eye on.

On top of those, about a dozen other states (New York, Washington, Oregon, and Georgia, to name a few) have passed chatbot laws in the past year or so. Most target AI companion apps, the kind people use like a friend or a therapist, not a small business answering its own phone. I won’t spend time on them here, but you should know they exist because they all point in the same direction.

Every State at a Glance: AI Disclosure and Call Recording Laws

Here’s a map that puts both issues side by side: the AI disclosure laws above and the older call recording laws that have been on the books for decades. Hover over any state for the short version, and click it to read the actual statute on the state’s own website (or the closest thing to it, since a few states only publish their laws through LexisNexis).

• Blue: an AI disclosure law that can reach a business using an AI agent. Colorado’s doesn’t kick in until January 1, 2027.
• Orange: you need the caller’s consent (or an up-front “this call is recorded” announcement) to record the call.
• Red: both.
• Grey: neither, as far as a business answering its own phone is concerned. A dozen or so grey states have passed narrower AI laws aimed at companion apps or AI therapy, and I note those in the hover text.

This map is my best reading of the laws as of October 2026, not legal advice. Laws change, so click through to the source before you rely on it.

What About States With No AI Law?

Even in states with no AI law, the Federal Trade Commission and several state attorneys general have already said that a bot pretending to be human is a deceptive business practice under decades-old consumer protection laws.

So these laws are all a little different, but when you line them up side by side, they’re asking for the same three things:

1. The AI should say it’s an AI right at the start.
2. It should never claim to be a human.
3. If someone asks whether they’re talking to a real person, it has to tell the truth.

The One-Sentence Fix for Your AI Voice Agent

Before I get into this, one important note: everything here assumes you’re using an AI agent to answer incoming calls and reply to incoming messages. In other words, the seller reached out to you first. Having an AI make outgoing calls or send outgoing texts is a totally different situation with a different set of rules. I’ll cover that further down.

For incoming calls, this is the part I think is kind of great, because one sentence takes care of most of it. Here’s the first thing my voice AI agent says when it picks up:

“Thanks for calling Seth’s Land Company. This is Sam, Seth’s AI assistant, on a recorded line. How can I help?”

That’s it.

The words “AI assistant” do the heavy lifting because the caller hears them before you say a word. That covers the “tell them at the start” part of every one of these laws.

If you want a template, here’s the version I put in the video description:

“Thanks for calling [Company]. This is [Name], our AI assistant, on a recorded line. Are you calling about selling your land?”

You want the AI agent itself to say this right at the start of the call, because that’s how it gets captured in the recording. If anybody ever questions it later, the proof is sitting right there.

Why “On a Recorded Line” Is in There

That opening sentence is actually doing two jobs.

First, you really do want recordings of these calls. In Stride CRM, you can automatically record and transcribe every call (or not, it’s your choice). But if you’re going to have an AI agent answer your phone, you might as well record the calls. If a seller comes back six months later and claims your AI told them it was a real person, or claims it never said it was an AI, the recording shows exactly what both sides said.

Beyond how useful recordings and transcripts are, that’s a strong asset to have in your back pocket. But it only works if the disclosure is inside the recording, which is why the agent should say it at the very beginning.

Second, recording a call is its own legal issue. About a dozen states (California being a big one) require that everyone on the call know it’s being recorded. California, in particular, lets a person sue for $5,000 per call if they weren’t told. Those laws have been around for decades, and I won’t go deep on them here. But since you have no idea where a caller is sitting when they call you, the easy move is to have your agent say “on a recorded line” on every call.

So that one sentence tells the caller it’s an AI, tells them the call is recorded, and gives you proof that it did both. All in about four seconds.

The Paragraph to Add to Your AI Agent’s Prompt

The opening sentence covers the start of the call. But what about 10 minutes in, when a seller stops and says, “Wait a minute, am I talking to a real person?”

For extra insurance, I’d add one paragraph to your agent’s instructions (some people call this the “prompt,” the written set of directions that tells the agent how to behave). In plain English, it says the agent is an AI and should never say or hint that it’s human. If anyone asks whether they’re talking to a real person, it tells the truth right away and offers to have a person call them back. And if somebody says “stop calling me” or “take me off your list,” it stops.

Here’s the exact wording I use. It’s free, so feel free to copy and paste it into your own agent:

You are an AI assistant, not a human. Never say or imply that you are a human, a real person, or a licensed professional, even if asked to role-play. If anyone asks in any form whether they are talking to a human, a real person, a bot, a robot, a recording, or AI, answer right away: “I’m an AI assistant for [Company]. I can keep helping, or have a person from our team call you back. Which would you prefer?” Do this every time it comes up. Never claim personal experiences, a location, or anything only a human would have. If the caller objects to talking with an AI or to being recorded, offer a callback from a person, collect the best number and time, and end the call. If the caller says stop calling, remove me, take me off your list, or anything similar, confirm once that they won’t be contacted again, tag the contact Do Not Contact, and end the call.

Does This Apply to Text Messages and Website Chat?

Yes. This applies to your chat AI agent too.

If you’ve got an AI replying to text messages, running the chat box on your website, or answering people on Facebook Messenger or WhatsApp, the same rules apply. (Again, I’m talking about the AI replying to people who messaged you first, not the AI sending out texts on its own.) In fact, New Jersey and California laws were written for online bots before voice AI agents were even a thing.

The fix is really similar:

• The first message your chat agent sends should say what it is. Something like, “Hi, this is Sam, the AI assistant for Seth’s Land Company.”
• If the conversation stops and picks back up a day or two later, it should say it again.
• If you have a chat box on your website, put one line right above where people type that tells them they’re chatting with an AI assistant.
• Paste a chat version of the paragraph into your chat agent’s instructions.

Here’s the chat version:

You are an AI assistant, not a human. Your first message in every new conversation must say so: “Hi, this is [Name], the AI assistant for [Company].” If a conversation picks back up on a new day, say it again in your first reply. Never say or imply that you are a human, a real person, or a licensed professional, even if asked to role-play. If anyone asks in any form whether they are talking to a human, a real person, a bot, or AI, answer right away: “I’m an AI assistant for [Company]. I can keep helping, or have a person from our team reach out. Which would you prefer?” Do this every time it comes up. Never claim personal experiences, a location, or anything only a human would have. If the person objects to chatting with an AI, offer to have a person reach out and stop messaging. If the person says stop, unsubscribe, remove me, leave me alone, or anything similar, confirm once that they won’t be contacted again, tag the contact Do Not Contact, and stop messaging.

And here’s the line I’d put above your website chat box:

You’re chatting with an AI assistant. Chats are saved and processed by our service providers. See our Privacy Policy.

What About Using AI to Make Outbound Calls?

This is separate from disclosure laws, and it applies when the AI dials the phone or sends the first text.

In 2024, the FCC (the federal agency that regulates phone calls) said an AI voice counts as an “artificial voice” under the federal robocall law. That basically means you can’t have an AI call someone’s cell phone unless they gave you permission ahead of time. The penalty is anywhere from $500 to $1,500 per call.

For now, I’m only letting AI answer my calls. I don’t have any immediate plans to let it make them. Who knows, a lot could change in the next five to 10 years. And I’m not saying you can never do it, because with written permission, there’s a legal way. But if you want to go down that road, I’d talk to an attorney first. The law on this is very, very new, and that’s just not a risk I want to take with my own business right now.

How to Keep Up With AI Laws (Without Making It a Part-Time Job)

Most of the laws I just walked through didn’t exist two years ago, and there are bills sitting in a bunch of state legislatures right now. Here’s how I’d keep an eye on it:

1. Check the Future of Privacy Forum’s tracker. It’s a nonprofit that keeps a free, state-by-state tracker of chatbot laws. Here’s the link: Future of Privacy Forum Chatbot Legislation Tracker.
2. Do a quick search once or twice a year. Search for “[your state] AI chatbot disclosure law” and see what comes up. Big law firms publish free summaries whenever a new law passes.

That’s really all it takes to stay ahead.

I can’t tell you this makes you bulletproof, because some of it is still being sorted out in the courts. But from everything I’ve read, that one sentence and that one paragraph cover most of the problems you’d likely run into.

Setting This Up in Stride CRM

You can set this up on pretty much any AI agent, and that one sentence works the same way on all of them.

But if you want one that was built specifically for land investors, that’s what we put together at Stride CRM. In case you didn’t know, I’m one of Stride’s co-founders, so I’m obviously a little biased. Take that for what it’s worth.

Here’s how it works:

• You can get a voice AI agent that answers your phone calls and a chat agent that handles your text messages and website chats.
• Both put the information they collect directly into your CRM.
• The intake questions for land sellers, plus the sentence and paragraph above, are already loaded in when you sign up, so you’re not starting from a blank page.

You can get this with the full Stride CRM package, where your direct mail goes out through Stride Mailer Studio, calls come in, the agent answers them, and everything lives in one place.

Or, if you only want AI agents without the full CRM, you can get them separately at Stride Agents.

And if none of this is for you, that’s totally cool too. I hope this takes a little of the mystery out of the legal side of using AI in your business.