AI for Law Firms

AI Receptionist Disclosure Rules by State: What Your Law Firm Bot Must Say

AI Receptionist Disclosure Rules by State: What Your Law Firm Bot Must Say

We may earn money or products from the companies mentioned in this post.

Disclosure: This article contains affiliate links. If you sign up for a product through one of these links, AI for Law Firms may earn a commission at no extra cost to you.

If your firm uses an AI receptionist, whether it answers calls, runs a web chat, or calls leads back, the first thing it says matters a great deal. There’s no single national rule, but the pieces fit together: bar ethics opinions on misleading communications, state consumer-protection laws on AI chatbots, federal robocall rules, and recording-consent statutes. This guide walks through what each says, state by state where it varies, and gives you an opening script that meets the strictest of them.

Not legal advice. This is general information as of October 2026. Several of these laws are new or pending. Confirm the current text in every state where your callers are located.

The short answer

Have your AI receptionist say, in its first sentence, that it’s an AI and not a lawyer or human staff member. Tell callers whether the call is recorded or transcribed, and get consent before it starts. Offer a quick way to reach a person. Don’t let it give legal advice. That one design meets or exceeds every rule below.

Bar ethics opinions

Florida: Opinion 24-1

Florida is the most direct. Its 2024 opinion says generative AI chatbots that talk with clients or third parties “must include a disclaimer indicating that the chatbot is an AI program and not a lawyer or employee of the law firm,” and that “a lawyer must inform prospective clients that they are communicating with an AI program and not with a lawyer or law firm employee.” It also warns against “an overly welcoming generative AI chatbot that may provide legal advice, fail to immediately identify itself as a chatbot, or fail to include clear and reasonably understandable disclaimers” (Florida Bar Ethics Opinion 24-1).

Oregon: Formal Opinion 2026-208

Oregon’s 2026 opinion on chatbots and AI agents answers whether lawyers may use them for intake with “Yes, qualified,” as reported by NWSidebar. The version presented to the OSB Board of Governors notes that “more than one state has opined that the use of a chatbot that does not identify itself as a chatbot rather than a person could be a misleading communication” under RPC 7.1. It also covers vendor diligence, supervision, and duties to prospective clients (OSB Board of Governors materials). Check the final published opinion on the OSB website for exact wording.

Everywhere else: Rules 7.1 and 1.18

Even without a specific opinion, Model Rule 7.1 bars misleading communications about a lawyer’s services. A human-sounding bot that lets a caller believe they’re talking to your paralegal is an obvious risk. Model Rule 1.18 makes information from a prospective client confidential even if they never hire you. That matters a great deal when software is asking the questions. See our AI ethics rules by state hub for your jurisdiction.

State AI disclosure laws

Utah: Utah Code § 13-77-103

Utah’s law, as amended by SB 226 (effective May 7, 2025), works on two levels (Utah Code § 13-77-103):

  • Consumer transactions: a supplier using generative AI must disclose that the person is interacting with AI “and not a human, if the individual asks or otherwise prompts the supplier about whether artificial intelligence is being used.”
  • Regulated occupations (which include lawyers): providers must “prominently disclose” generative AI use in the provision of regulated services when it “constitutes a high-risk artificial intelligence interaction.” The disclosure is made “verbally at the start of a verbal interaction” and “in writing before the start of a written interaction.”

The enrolled bill also has a safe harbor for AI that clearly and conspicuously discloses “at the outset of any interaction” and “throughout the interaction” that it is generative AI and not human (SB 226 enrolled text). For a law firm, disclosing up front is the simple way to qualify.

Maine: 10 M.R.S. § 1500-DD

Maine’s law says 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.” A violation is a violation of the Maine Unfair Trade Practices Act (10 M.R.S. § 1500-DD; enacted as P.L. 2025, ch. 294, approved June 12, 2025). The definition of “artificial intelligence chatbot” covers both “textual or aural communications,” so voice receptionists are included.

New Jersey: proposed bills only

New Jersey has no enacted general AI-disclosure law that we could find. Two 2026 bills are pending:

  • A4730 would require “a clear and conspicuous verbal or written notice at the beginning of the interaction” when generative AI interacts with a consumer in a way that could make them believe they’re talking to a human (A4730).
  • S3668 would require online platforms to notify consumers that “an artificial intelligence system is communicating with the consumer” and tell them how to reach a human (S3668).

Watch their status. A design that already discloses up front and offers a human contact would likely satisfy both if they pass.

California: outbound calls (Public Utilities Code § 2874)

California’s AB 2905, effective January 1, 2025, amended the state’s rules for automatic dialing-announcing devices. Before the device runs, an “unrecorded, natural voice announcement” must state the nature of the call and the caller’s business details, ask whether the person consents to hear the prerecorded message, and “inform the person called if the prerecorded message uses an artificial voice.” “Artificial voice” means one “generated or significantly altered using artificial intelligence” (Cal. Pub. Util. Code § 2874). This mainly matters for outbound callbacks and reminders.

Federal: the TCPA and AI voices

In a February 2024 Declaratory Ruling, the FCC confirmed that the TCPA’s restrictions on “artificial or prerecorded voice” calls “encompass current AI technologies that generate human voices.” As a result, those calls “require the prior express consent of the called party” absent an emergency or exemption. Calls that include telemarketing require prior express written consent (FCC 24-17).

So an AI that automatically calls back every web lead’s cell phone needs documented consent. The TCPA gives a private right of action (47 U.S.C. § 227), and damages add up per call.

Recording and transcription consent is a separate question

Telling a caller they’re talking to an AI doesn’t satisfy recording law. Some states require every party to consent before a call is recorded. California (Penal Code § 632) and Florida (Fla. Stat. § 934.03) are two commonly cited examples. Others require only one party’s consent. If callers may be in an all-party-consent state, the safe approach is to get consent from everyone. Treat AI transcription as recording.

State-by-state summary

Jurisdiction Source What it requires Status
Florida Bar Ethics Op. 24-1 Chatbot must disclose it’s AI, not a lawyer or employee Advisory ethics opinion
Oregon Formal Op. 2026-208 A non-identifying chatbot may be misleading under RPC 7.1; supervise the vendor Ethics opinion (2026)
Utah Utah Code § 13-77-103 Disclose when asked; regulated occupations must disclose up front for high-risk interactions In force (May 7, 2025)
Maine 10 M.R.S. § 1500-DD Clear and conspicuous notice if a consumer could think it’s a human In force
New Jersey A4730, S3668 Up-front AI notice; S3668 adds human-contact info Proposed only
California Pub. Util. Code § 2874 Outbound automated calls must disclose an artificial voice In force (Jan 1, 2025)
Federal TCPA / FCC 24-17 Prior express consent for AI-voice calls In force

Opening scripts that work

Inbound call, recorded:

“Thanks for calling Smith Law. I’m an AI receptionist, not a lawyer or a person on our staff. I can take your information, schedule a consultation, or connect you to our team. This call is recorded and transcribed. Is that okay with you?”

Inbound call, not recorded:

“You’ve reached Smith Law’s AI receptionist. I’m not a lawyer and can’t give legal advice. I can schedule a call or connect you with a person. Just say ‘person’ at any time.”

Outbound callback (with documented consent):

“Hi, this is an AI voice calling for Smith Law, returning your inquiry from today. You can say ‘stop’ to opt out of automated calls.”

Web chat banner:

“You’re chatting with an AI assistant. It’s not a lawyer, can’t give legal advice, and doesn’t create an attorney-client relationship. A member of our team reviews every inquiry.”

Most chatbot builders let you set this as the bot’s first message. In Chatbase, for example, you can set it as the chat bubble’s initial message, add a short footer disclaimer, and choose which content the bot is trained on.

What the bot should and shouldn’t do

Fine: office hours, directions, scheduling, callback requests, basic contact details, the general type of matter, and the opposing party’s name for a preliminary conflict screen.

Not fine: assessing whether the caller has a case, quoting fees, explaining deadlines or statutes of limitations, recommending actions, or accepting or rejecting a client without a human. Oregon’s opinion also cautions that autonomous actions, such as sending an engagement letter, should require lawyer approval.

Collect as little information as possible. Every detail the bot gathers becomes prospective-client information under Rule 1.18, and you have to protect it and enter it into conflicts. Build an emergency escalation for words like “arrest,” “eviction,” “domestic violence,” or “self-harm” that routes the call straight to a person or to emergency resources. See what intake you can (and can’t) automate and our roundup of lawyer AI chatbots.

Vendor diligence

Your receptionist vendor processes prospective-client confidences for you. Before you sign, get written answers on data retention, human access to recordings, model training, subcontractors (many receptionist products run on another company’s AI model), breach notice, and deletion when the contract ends. Oregon’s opinion specifically notes that a chatbot vendor “may be utilizing other businesses’ software as its foundation,” which means your diligence has to cover those companies too.

Frequently asked questions

Does my AI receptionist have to say it’s AI?

In Florida, yes, under Opinion 24-1. In Utah and Maine, state statutes can require it. Everywhere else, Rule 7.1’s ban on misleading communications makes up-front disclosure the safe choice.

Does New Jersey require AI disclosure?

Not yet, as far as we can find. A4730 and S3668 are proposed bills. Design as if they will pass.

Is AI transcription the same as recording?

Treat it as recording. Get consent before transcription starts, especially if callers may be in an all-party-consent state.

Can my AI receptionist call leads back?

Only with proper consent. The FCC treats AI-generated voices as “artificial” voices under the TCPA.

Key takeaways

  • Disclose AI identity in the first sentence, and in writing before any chat starts.
  • Florida, Oregon, Utah, and Maine all point the same way. New Jersey’s bills are proposed only.
  • Recording consent and TCPA consent are separate obligations.
  • Keep the bot administrative, collect little, and escalate to a human quickly.

Sources