Home / AI chatbot disclosure

Guide for mortgage brokers

Do you have to tell borrowers they are talking to AI?

Last reviewed: November 9, 2026. General information, not legal advice. State rules differ and change; confirm how they apply to your tools and your borrowers' states with counsel.

The short answer

There is no single federal rule that requires a label on every AI chat. But several states do require disclosure in specific situations, and letting a borrower believe a bot is a person is risky everywhere. For a licensed mortgage business, the simplest safe practice is to say it is AI at the start of every automated conversation, and say so again whenever someone asks.

That one habit addresses the four state laws below and costs little to set up. The rest of this page shows where the rules come from, where AI is quietly talking to your borrowers, and scripts you can copy.

State AI chatbot disclosure rules that can reach mortgage brokers, as of November 9, 2026
StateLawStatusWhat it requires
UtahUtah Code 13-77-103In effect (current version effective May 7, 2025)Two duties: answer honestly when a consumer asks if it is AI, and, for licensed occupations, disclose up front in high-risk interactions such as collecting financial data or giving financial advice.
Maine10 M.R.S. section 1500-DD (Public Law 2025, chapter 294)In effect since September 24, 2025A chatbot that may lead a reasonable consumer to think they are talking to a person needs a clear and conspicuous notice. A violation is a violation of the Maine Unfair Trade Practices Act.
CaliforniaBusiness and Professions Code 17940 to 17943 (the "bot" law)Operative since July 1, 2019Unlawful to use a bot online to mislead a person in California about its artificial identity to drive a sale. Disclosing that it is a bot avoids liability.
ColoradoHB 26-1263, Conversational AI Service Operator Requirements (C.R.S. 6-1-1708)Duties apply on and after January 1, 2027Operators of public conversational AI services must disclose AI at the start of each day of use, every three hours or persistently, and when asked. Customer service and transactional tools are excluded from the definition.

Utah: disclose when asked, and up front for licensed work

"A supplier that uses generative artificial intelligence to interact with an individual in connection with a consumer transaction shall disclose to the individual that the individual is interacting with generative artificial intelligence and not a human, if the individual asks or otherwise prompts the supplier about whether artificial intelligence is being used."Utah Code 13-77-103(1)(a)

In plain English: If a borrower asks whether they are talking to AI, the answer must be yes.

"prominently disclose when an individual receiving services is interacting with generative artificial intelligence in the provision of regulated services if the use of generative artificial intelligence constitutes a high-risk artificial intelligence interaction"Utah Code 13-77-103(2)(a)

In plain English: People in licensed occupations must disclose AI up front, without being asked, in high-risk interactions.

Chapter 77 defines a high-risk interaction to include collecting sensitive personal information, including financial data, and giving personalized financial advice or services. Utah mortgage loan originators are licensed through the Utah Department of Commerce, so an AI assistant that collects income or asset details from Utah borrowers is likely to need the up-front disclosure. See state AI laws that affect mortgage brokers for the Utah details.

Maine: no bot that passes as a person

"A person may not use an artificial intelligence chatbot or any other computer technology to engage in trade and commerce with a consumer 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 notified in a clear and conspicuous manner that the consumer is not engaging with a human being."10 M.R.S. section 1500-DD(2)

In plain English: If your chat or text tool could pass for a person, tell Maine borrowers clearly that it is not.

The law defines an artificial intelligence chatbot as a software application, web interface or computer program that simulates human conversation through text or voice. The Maine Bureau of Consumer Credit Protection flagged this law to its licensees in a September 24, 2025 advisory, the day it took effect.

California: the bot disclosure law

"It shall be unlawful for any person to use a bot to communicate or interact with another person in California online, with the intent to mislead the other person about its artificial identity for the purpose of knowingly deceiving the person about the content of the communication in order to incentivize a purchase or sale of goods or services in a commercial transaction or to influence a vote in an election. A person using a bot shall not be liable under this section if the person discloses that it is a bot."Cal. Bus. and Prof. Code 17941(a)

In plain English: Online bots that hide that they are bots to drive a sale are unlawful in California. A clear disclosure is the safe harbor.

Section 17941(b) says the disclosure must be clear, conspicuous and reasonably designed to inform the people the bot talks with that it is a bot.

Colorado: chatbot rules start January 1, 2027

Colorado's HB 26-1263, signed May 29, 2026, adds duties for operators of a "conversational artificial intelligence service" starting January 1, 2027. It is separate from Colorado's ADMT law for lending decisions, SB 26-189.

"Be provided at the beginning of a user's first interaction with a conversational artificial intelligence service for each day of interaction;"C.R.S. 6-1-1708(3)(a)

In plain English: Covered operators must say it is AI at the start of each day's first conversation, at least every three hours or with a persistent label, and whenever a user asks.

"Is primarily designed to provide commerce-related or transactional assistance, including product or service recommendations, shopping, ordering, payments, delivery, returns, customer support, or customer service;"C.R.S. 6-1-1701(3.5)(b)(II), excluded from the definition

In plain English: A chat tool built for customer service or transactions is outside the definition, so many mortgage website chat tools may not be covered.

The definition also excludes tools used only inside a business, and AI features inside other software that are not designed for emotional companionship. A general-purpose AI assistant offered to the public is more likely to be covered. Ask counsel how your tool fits, and watch the Colorado Attorney General's rules (4 CCR 904-6), which were still in draft at our last review. Adding the disclosure anyway costs little and covers the other states.

Everywhere: deception and calling rules

The FTC Act prohibits unfair or deceptive acts or practices in or affecting commerce (15 U.S.C. 45(a)(1)), and many states have similar laws. A bot that claims to be a person, or claims a loan officer reviewed something they did not, is the kind of statement those laws reach. If your AI places calls or sends texts, the Telephone Consumer Protection Act also applies: the FCC confirmed in 2024 that AI-generated voices are "artificial" voices under the TCPA. See AI voice agents and AI texts to leads.

Where AI may already be talking to your borrowers

  • 1. Website chat widget. Often the first place to look. If it answers with AI, label it in the widget header, not only in your privacy policy.
  • 2. AI text follow-up on new leads. An AI that texts a lead back in seconds and keeps the conversation going is a chatbot by any of these definitions. Calling and texting rules also apply: see AI voice agents and AI texts to leads.
  • 3. AI phone receptionist or voice agent. A voice that answers or places calls for you. Say it is an AI assistant in the first sentence.
  • 4. Social media and messaging inbox replies. Auto-replies that hold a conversation in direct messages count too.
  • 5. Borrower portal assistant. Many point of sale systems and borrower portals now ship an AI helper. Check whether yours is on and what it tells borrowers.
  • 6. Emails drafted by AI and sent by a person. A loan officer who reviews and sends an AI draft is still the one talking. The chatbot rules above target automated conversations, but the content is still yours, so review it before it goes out.

Disclosure scripts you can copy

Chat widget header (persistent label)

AI assistant. Not a person. Ask for a loan officer anytime.

First chat or text message

Hi, I am an AI assistant for [Company name], not a person. I can answer general questions and collect a few details for your loan officer. A licensed loan officer reviews everything before you get any rate, terms or decision. Type "loan officer" anytime to reach a person.

Answer to "Am I talking to a real person?"

No, I am an AI assistant, not a person. Would you like me to connect you with a licensed loan officer now?

AI voice assistant greeting (inbound calls)

Thanks for calling [Company name]. I am an AI assistant, not a person. I can take a message or book a time with a loan officer. Say "loan officer" anytime to be transferred.

Handoff to a person

I am handing you to [Loan officer name], a licensed loan officer, NMLS [number]. They can see this conversation, so you will not need to repeat yourself.

Replace the bracketed parts before use. Keep the words "AI" and "not a person" in every version; those are what the Utah, Maine and California rules turn on.

Set it up right: eight steps

  • 1. List every tool that talks to borrowers. Put each one in your AI use inventory and mark it borrower-facing.
  • 2. Turn on a visible label. Use a persistent "AI assistant" label in the chat header and a first message that says it is AI. A persistent label is one of the options Colorado accepts. (C.R.S. 6-1-1708(3)(b))
  • 3. Script the "Am I talking to a real person?" answer. The bot must answer yes, it is AI, every time it is asked. Test it with different wordings. (Utah Code 13-77-103(1)(a))
  • 4. Give a fast path to a person. Offer "talk to a loan officer" in every conversation and route it to a licensed person.
  • 5. Keep the bot away from credit decisions and quotes. Do not let it tell a borrower they are approved or denied, or quote a rate or terms without loan officer review. See the AI marketing fair lending checklist.
  • 6. Check the vendor's settings, not just its marketing. Ask whether the AI disclosure can be turned off by staff, whether transcripts are kept and where, and whether the vendor trains on your conversations. Use the AI vendor due diligence questions.
  • 7. Keep transcripts with the loan file rules. Chat and text transcripts can hold borrower data. Store them under your record retention and security program.
  • 8. Review every quarter. Vendors add AI features often. Recheck the label, the scripts and this page's state list each quarter.

Need a written rule for this? Add an "AI that talks to borrowers" section to your AI policy with the mortgage AI policy template.

Get your team to sign off on how AI talks to borrowers

The free BrokerLumen portal plan lets you publish your AI rules, have up to 10 staff acknowledge them and see who has not signed yet. Start free, then add a kit if you want the full policy, training deck and vendor questionnaires.

Start freeCompare kits

Frequently asked questions

Do mortgage brokers have to disclose that a chatbot is AI?

In some states, yes. Utah requires disclosure when a consumer asks and, for licensed occupations, up front in high-risk interactions such as collecting financial data. Maine requires a clear notice when a chatbot could mislead a reasonable consumer into thinking it is human. California prohibits online bots that hide their artificial identity to drive a sale. Disclosing at the start of every automated conversation is the simplest way to cover all of them.

Does Colorado's chatbot law apply to a mortgage website chat?

Colorado HB 26-1263 applies to operators of conversational AI services from January 1, 2027, but its definition excludes tools primarily designed for customer support, customer service and other transactional assistance, and tools used only inside a business. Many mortgage website chat tools may fall in that exclusion. A general-purpose AI assistant offered to the public is more likely to be covered. Ask counsel how your tool fits.

What should a mortgage chatbot say when someone asks if it is a person?

It should say clearly that it is an AI assistant and not a person, and offer to connect the borrower with a licensed loan officer. Utah Code 13-77-103(1)(a) requires that disclosure when a consumer asks.

Is an AI-drafted email that a loan officer sends a chatbot?

Generally no. The chatbot rules target automated conversations. When a loan officer reviews and sends an AI draft, the loan officer is communicating, and is responsible for the content.

Do these rules apply to AI phone calls?

The state chatbot laws cover voice as well as text in several cases, and outbound AI voice calls are also covered by the Telephone Consumer Protection Act. The FCC confirmed in February 2024 that AI-generated voices are artificial voices under the TCPA, so those calls need prior consent.