Is my business liable for what our AI chatbot says?
Yes. The leading case, Moffatt v. Air Canada (2024), held an airline to a refund rule its website chatbot invented, and rejected the argument that the chatbot was a separate entity responsible for itself. The case was a small Canadian tribunal ruling and is not binding in the United States, but the principle is the one U.S. courts and regulators apply: what your website, bot or AI receptionist says is your statement. Insurance may respond through professional liability, but only if the wording does not exclude AI and the loss is the kind it covers.
Tell us the situation.
One licensed human replies the same business day — not five agents, not an auto-responder. If the policy you have is already the right one, we will tell you.
We use this to quote and service your insurance, and we do not sell it or pass it to lead networks. Privacy policy.
Send us what you have. We’ll tell you if it’s right.
Your current policy paperwork, a renewal notice, or nothing at all. A licensed agent reads it and replies the same business day.
What the Air Canada chatbot case decided
A traveler asked the airline’s website chatbot about bereavement fares and was told he could apply for the discount after his flight. The airline’s actual policy did not allow that. He bought the ticket, applied, and was refused. The tribunal sided with him and awarded about $812 Canadian in total.
The airline argued, in effect, that the chatbot was its own entity and the airline was not responsible for what it said. The tribunal rejected that, and the commentary that followed focused on one idea: a company is responsible for the information on its website whether a person typed it, a static page displayed it or a bot generated it.
Be careful with the limits. It was a small-claims tribunal in British Columbia, so no U.S. court has to follow it, and it is not even clear the bot used generative AI. Its value to an owner is that it is the first widely reported ruling on the question, and it went the way most lawyers expected.
Does a disclaimer on the chat window protect you?
A line saying “this assistant may make mistakes” is good practice and may help with an angry customer, but it is not a shield. The airline made a version of that argument, that the customer should have checked the policy page himself, and it did not carry the day. A disclaimer is most credible when the bot is also built not to promise things: no prices, refunds, warranties or coverage statements that have not been approved.
The better protection is a design one. Let the bot answer from your own published pages, send anything about money or promises to a person, and keep the transcript so you can see what was said.
The other risk in a chat or phone bot: recording and disclosure
A chat widget or AI phone assistant usually records the conversation and passes it to the vendor that runs the tool. Plaintiffs have argued that this is eavesdropping under state wiretapping laws, particularly California’s, and in 2025 a federal court let a class action against an AI phone-ordering vendor continue after a motion to dismiss. That is an allegation moving forward, not a finding that anyone broke the law, but it is why the notice at the start of a chat or call matters.
If the AI calls people rather than answers them, the FCC ruled in February 2024 that an AI-generated voice is an artificial voice under the Telephone Consumer Protection Act. A call using one needs the person’s prior consent, the same as a prerecorded message.
Some states also require you to say you are a bot. Utah requires disclosure when asked and, for licensed professions, up front. California has a law against using a bot to mislead someone about its identity to make a sale. Check the states where your customers are.
Which insurance responds when a bot gets it wrong
A customer who lost money because your bot gave wrong information is usually claiming financial loss from a mistake in your service. That is the territory of professional liability, also called errors and omissions (E&O). Three things decide whether yours responds: whether your business is within the professional services the policy describes, whether the policy carries an AI exclusion, and whether the claim is for a mistake or for a promise, because many policies do not pay to honor a promise or to refund a fee.
A privacy claim about the recording sits on cyber liability or sometimes media liability, and exclusions for wiretapping and call-recording statutes are common in both. Ask about them by name.
General liability is the wrong place to look for a wrong answer, and since January 2026 it may also carry a generative-AI exclusion.
What a chatbot claim looks like, and where to look
Bot quoted a price or refund you do not offer
- Likely policy
- Professional liability / E&O
- Check in the wording
- Is a promise or refund excluded? Is your service a covered professional service?
Bot gave wrong advice a customer acted on
- Likely policy
- Professional liability / E&O
- Check in the wording
- Any exclusion for AI or for unlicensed advice
Class action over recording chats or calls
- Likely policy
- Cyber liability; sometimes media
- Check in the wording
- Exclusions for wiretapping, call recording or the TCPA
Bot called people without consent
- Likely policy
- Rarely insured; fines and statutory damages are often excluded
- Check in the wording
- TCPA exclusion; whether defense costs are included
Bot said something that injured or damaged
- Likely policy
- General liability
- Check in the wording
- Generative-AI exclusion; professional services exclusion
A map of where to look. Your own wording decides.
The detail
In Moffatt v. Air Canada (2024 BCCRT 149, decided February 14, 2024) a British Columbia tribunal held the airline responsible for wrong refund information its website chatbot gave a customer, rejected the argument that the chatbot was a separate entity answerable for itself, and awarded about $812 Canadian in total.
Commentators note Moffatt was a small-claims tribunal decision in Canada, not binding on U.S. courts, and that it is unclear whether the airline’s bot used generative AI at all; the principle they draw is that a business answers for what its website says, however it was produced.
McMillan LLP, “A Word of Caution: Company Liable for Misrepresentations Made by Chatbot”
In Mata v. Avianca a federal judge in New York fined two lawyers and their firm $5,000 on June 22, 2023 after they filed a brief citing cases ChatGPT had invented, writing that existing rules give lawyers a gatekeeping role to ensure the accuracy of their filings.
On August 11, 2025 a federal court in the Northern District of California let a class action proceed against the maker of an AI phone-ordering assistant, Taylor v. ConverseNow, alleging its recording of a customer’s call to a pizza restaurant violated California’s wiretapping and call-recording laws; it is an allegation that survived a motion to dismiss, not a finding of liability.
On February 8, 2024 the FCC ruled unanimously that AI-generated voices count as “artificial” voices under the Telephone Consumer Protection Act, so a call using one needs the recipient’s prior consent; the ruling clarified existing rules rather than banning the technology.
Cooley, “FCC: AI-Generated Robocalls Illegal Under the TCPA” (February 15, 2024)
Utah’s AI Policy Act, effective May 1, 2024, requires a business using generative AI to talk to consumers to disclose that the consumer is not dealing with a human when asked, and licensed professions must disclose without being asked; one law-firm source notes it was later amended and is scheduled to be repealed in July 2027.
Mayer Brown, “Utah Enacts AI-Focused Consumer Protection Bill” (May 13, 2024)
On September 25, 2024 the FTC announced Operation AI Comply, five enforcement actions against businesses using AI to deceive; its chair said there is no AI exemption from the laws on the books, and one company, DoNotPay, agreed to pay $193,000 and stop claiming its “AI lawyer” could replace professional services without proof.
What this means for you
- Treat every bot answer as published by you. Read a sample of real transcripts monthly. The wrong answer you find yourself costs nothing.
- Fence what the bot may promise. Prices, refunds, warranties, coverage and deadlines go to a person or to a page you control.
- Say it is a bot and say it is recorded. Put it in the first message of the chat and the opening of the call. It is cheap and some states require it.
- Ask your agent three questions in writing. Does my professional liability exclude AI? Does my cyber or media policy exclude wiretapping claims? Does my general liability carry CG 40 47 or CG 40 48?
- Check the vendor contract. Who owns the transcripts, can the vendor train on them, and does the vendor indemnify you for anything? Most say little.
Related questions
- Did Air Canada have to pay because the chatbot was AI?
The tribunal’s reasoning did not depend on the technology. It held the airline responsible for information on its website. Commentators note it is unclear whether the bot used generative AI, which is why the lesson applies equally to a scripted bot, a generative bot and an AI receptionist.
- Does a “not legal advice” or “may be inaccurate” notice protect me?
It helps, but it does not replace controlling what the bot says. A notice does not undo a specific promise a customer relied on, and it does not answer a recording or disclosure claim.
- Is an AI receptionist different from a website chatbot?
The liability for what it says is the same. The extra risks are call recording and consent laws, which vary by state, and the rules on AI voices in outbound calls.
- Will my professional liability policy pay if my bot gives wrong advice?
Possibly, if your business is one the policy describes and it has no AI exclusion, but we cannot say yes without your wording. Insurance companies are adding AI exclusions to some professional liability forms, so ask for the endorsements list.
- What professional liability covers
- What should I do if my bot already gave a customer wrong information?
Correct it with the customer in writing, fix the source the bot used, save the transcript and tell your insurance agent before anyone makes a demand, because many policies require prompt notice of a circumstance that may become a claim.
Need this handled?
Reading about it and having it handled are different jobs, and the second one is ours. We place the accounts other brokers decline — claims history, a cancellation, an unusual operation, a business nobody wants to underwrite twice — and we will tell you when the policy you already have is the right one. Tell us the situation and a licensed human replies the same business day.
Written by Kevin Kelsey, founder and principal agent, CA License #6011822, at Boring Insurance. Last updated 2026-10-25. See all guides.