Are You Liable for What Your AI Receptionist Says? (2026)

If your AI receptionist tells a caller there's no trip charge after hours, quotes a price you don't charge, or promises a technician at 7 AM, expect to be treated as if you said it yourself. The decisions so far treat a customer-facing AI as the business speaking, not as a separate actor that takes the blame. This is general information, not legal advice. Talk to a lawyer licensed where you operate about your own situation.

The rule that has always applied to your office staff now applies to your software: what it says on your behalf is on you.

What happened in Moffatt v. Air Canada?

In November 2022, after Jake Moffatt's grandmother died, Moffatt used the chatbot on Air Canada's website to ask about bereavement fares. The chatbot said a bereavement fare could be claimed retroactively, after the tickets were bought. Moffatt booked at the regular fare on that basis. When the refund request came in, Air Canada said its policy didn't allow retroactive claims.

Moffatt took the airline to British Columbia's Civil Resolution Tribunal. On February 14, 2024, the tribunal found Air Canada liable for negligent misrepresentation in Moffatt v. Air Canada, 2024 BCCRT 149.

Two parts of the decision matter for any business that lets AI talk to its customers:

The "separate legal entity" defense failed. Air Canada argued it couldn't be held liable for information given by its agents or representatives, including the chatbot. The tribunal called that "a remarkable submission" and rejected it.
The channel doesn't matter. The tribunal said it should be obvious to Air Canada that it is responsible for all the information on its website, and that it makes no difference whether that information comes from a static page or a chatbot.

The money was small: CAD $812.02 in total, made up of $650.88 in damages plus pre-judgment interest and tribunal fees. The principle isn't. Swap the fare for your after-hours trip charge, and a homeowner who relied on what your AI said has Moffatt's argument.

Has any other court ruled on AI chatbot liability?

Yes. On May 12, 2026, Germany's Higher Regional Court of Hamm (Oberlandesgericht Hamm) decided a case a consumer association brought against an aesthetic-treatment business, case 4 UKl 3/25. The company's website chatbot told users that its managing directors were specialist physicians in plastic and aesthetic surgery. They weren't, and some of the specialist titles the chatbot used don't exist.

The company argued the chatbot's answers weren't its own statements. The court disagreed. It described the chatbot as merely a technical means the company used to communicate with potential customers, one over which the company had sufficient control, and held the company responsible for its answers. The court allowed an appeal to Germany's Federal Court of Justice, so the final word may still change. Two decisions aren't a settled worldwide rule, but we haven't found a published decision going the other way.

Who is liable, my business or the AI vendor?

Toward the customer, both decisions point at the business the customer was dealing with. Whether you can recover any of that cost from your vendor depends on your contract, and software contracts, ours included, commonly cap the vendor's liability. Read the limitation-of-liability section before you sign.

Do AI disclosure laws apply to an AI receptionist answering your calls?

Mostly not the ones you see quoted. We read the primary sources. Here's what each one actually covers, as of September 2026.

California AB 2905 (2024)

AB 2905 amends Public Utilities Code section 2874, which governs automatic dialing-announcing devices: equipment that places outbound calls and plays a prerecorded message. It requires the live-voice announcement on those calls to tell the person called if the message uses an artificial voice. Chaptered on September 20, 2024, it's a rule for robocalls you place, not for calls your receptionist answers.

California SB 1001 (the bot disclosure law)

SB 1001, operative since July 1, 2019 (Business and Professions Code sections 17940 to 17943), covers bots it defines as automated online accounts on public-facing websites, applications and social networks. It targets knowingly misleading someone about a bot's artificial identity to push a sale or influence a vote. Its definitions don't name phone calls, though it's still a good reason never to let an AI claim to be human.

Colorado SB26-189

SB26-189, titled "Automated Decision-Making Technology," was signed on May 14, 2026, with requirements beginning January 1, 2027. It covers automated decision-making technology used in consequential decisions about a person's access to education, employment, housing, financial or lending services, insurance, health-care services, or essential government services. Setting up a furnace repair isn't one of those. Using AI to screen job applicants would be.

Utah SB 149 (2024) and SB 226 (2025)

Utah's law is the one here that does reach an ordinary business, narrowly. SB 149 created the state's AI disclosure rules in 2024, and SB 226, effective May 7, 2025, rewrote them:

A supplier using generative AI in a consumer transaction must disclose that the person is interacting with AI and not a human if the person asks, with a clear and unambiguous request to find out (Utah Code 13-75-103(1)).
Someone in a regulated occupation, one licensed or state-certified through Utah's Department of Commerce, must disclose AI use prominently, verbally at the start of a spoken interaction, but only in high-risk interactions: collecting health, financial or biometric data, or giving personalized financial, legal, medical or mental-health advice (13-75-101 and 13-75-103(2) and (3)).
It is not a defense to a violation of Utah's consumer protection statutes that generative AI made the statement (13-75-102).

The plain reading for a contractor: if a Utah caller clearly asks whether they're talking to a real person, the AI has to say it's an AI. If your receptionist collects card numbers or other financial data, ask a lawyer whether that makes the call "high-risk" for your licensed trade.

Texas HB 149 (TRAIGA)

The Texas Responsible Artificial Intelligence Governance Act, effective January 1, 2026, puts its AI disclosure duty on governmental agencies, which must tell consumers before or at the time of the interaction, and on health care providers using AI in a patient's service or treatment. It doesn't put that duty on private businesses generally, so it doesn't reach a Texas roofer's AI receptionist.

The FCC's AI voice ruling (FCC 24-17)

In a declaratory ruling adopted February 2 and released February 8, 2024, the FCC confirmed that the Telephone Consumer Protection Act's restrictions on "artificial or prerecorded voice" calls cover current AI technologies that generate human voices, so those calls generally need the called party's prior express consent. Those restrictions govern calls you make, not calls your customers place to you. They matter once you use AI for outbound calls such as appointment reminders or follow-ups: get consent first.

Recording laws apply either way

Recording is where an inbound receptionist, AI or human, runs into a real rule. Florida makes intercepting a call lawful when all parties have given prior consent (Fla. Stat. 934.03), and California prohibits recording a confidential communication without the consent of all parties (Cal. Penal Code 632). A short "this call may be recorded" at the start of the greeting is the usual way businesses handle it.

The takeaway: most of the new AI laws people cite don't govern your inbound line, but the old rules still do. Misrepresentation, consumer protection and recording consent apply to your AI exactly as they apply to a person on your staff.

What should a contractor do?

Going back to voicemail isn't the answer, since unanswered calls cost contractors real jobs (see our 2026 missed-call cost research). Run the AI with the controls you'd put on a new office hire:

Never let it invent prices or appointment times. If a price isn't in its rules, or it can't see your real schedule, it shouldn't guess.
Make "I'll have someone call you back" the default when it's unsure. A callback costs a few minutes. A wrong promise hands the customer the Moffatt argument.
Have it answer honestly when asked if it's an AI. Utah requires that when a caller clearly asks, and everywhere else a receptionist that lies about being human is a trust problem.
Keep recordings and transcripts, and review a sample every week. Ten calls a week catches a bad habit before a customer does, and the recording settles what was said.
Disclose recording where every party must consent, including Florida and California.
Write your emergency rules down explicitly: what counts as an emergency, who gets alerted, and what after-hours rate is stated before a tech goes out. Our after-hours answering service guide has a trade-by-trade emergency table and a template for after-hours rates.
Test it with seven calls before you trust it: an emergency, a routine evening caller, a request for a time it can't see, a request for a human, "Are you an AI?", a pricing question it shouldn't answer, and a returning customer. The script and scorecard are in how to test an AI receptionist.
Get consent before any outbound AI calls. Reminders and follow-ups fall under the FCC's 2024 ruling.

When it does get something wrong, as every receptionist eventually will, listen to the recording, make it right with the customer who relied on it, and fix the rule so it doesn't happen twice.

How LocalCall AI handles this

We built LocalCall AI for contractors, so these controls are part of setup:

It tells callers it's an AI when they ask.
Every call is recorded, with a transcript and summary in your portal, so you can review what was said and settle a dispute with the actual audio.
Rules are configured per business: your hours, what counts as an emergency in your trade, your after-hours rates, and what the AI may and may not say.
It doesn't invent quotes. If a price isn't in your rules, it takes the details for a callback.
Recording notices on request. If you operate where every party must consent, tell us at setup and we'll add a recording notice to your greeting.

None of that replaces a lawyer, and it doesn't take on your liability to your customers. It gives you the rules and the records. See pricing for what each plan includes.

FAQ

This is general information, not legal advice.

Is my business liable if my AI receptionist quotes the wrong price?

Very possibly. In Moffatt v. Air Canada (2024 BCCRT 149), a British Columbia tribunal held Air Canada liable for negligent misrepresentation after its chatbot gave wrong fare information, and a German court held a company responsible for its chatbot's false statements in May 2026. If a customer reasonably relied on a price your AI gave, expect to be treated as if your business quoted it.

Can a disclaimer that the AI might make mistakes protect my business?

Don't count on it. In Moffatt, the tribunal said Air Canada was responsible for all the information on its website, whether from a static page or a chatbot, and neither decision suggests a disclaimer alone solves the problem. The safer approach is no invented prices, no invented appointment times, and a callback when the AI is unsure.

Do I have to tell callers they're talking to an AI?

It depends where you operate. Utah requires a business using generative AI in a consumer transaction to say so when a person clearly asks, and requires up-front disclosure from licensed occupations in high-risk interactions. California AB 2905, Colorado SB26-189 and Texas HB 149 don't put that duty on a private contractor's inbound receptionist. Answering honestly when asked is the safe practice everywhere.

Is it legal to record calls answered by an AI receptionist?

The recording rules are the same whether a person or an AI answers. Florida and California require the consent of every party to the call, so businesses there typically play a short recording notice at the start. Keep the recordings and transcripts: they're your best evidence of what the AI said.

Does the FCC's 2024 AI voice ruling apply to AI receptionists?

FCC 24-17 confirmed that AI-generated voices count as "artificial" voices under the Telephone Consumer Protection Act, whose restrictions govern calls a business places. An AI answering calls customers make to you isn't placing them. If you use AI for outbound reminders or follow-ups, get prior express consent first.

Who is responsible, the AI vendor or my business?

Toward your customer, the decisions so far point at your business: Air Canada's argument that its chatbot was a separate legal entity was called "a remarkable submission" and rejected. Utah's statute also says AI making the statement is no defense to a consumer protection violation. Whether you can recover costs from your vendor depends on your contract.

Are You Liable for What Your AI Receptionist Tells a Customer? (2026)