Short answer
A chatbot hurts when it answers questions about policy, price or refunds from a source nobody keeps current and nobody checks. In Moffatt v. Air Canada (2024), a tribunal held the airline liable for its chatbot’s wrong answer on bereavement fares and rejected the idea that the bot was responsible for itself. Bots help on narrow, verified questions.
Key takeaways
- A British Columbia tribunal ordered Air Canada to pay $812.02 after its chatbot misstated the bereavement fare policy (Moffatt v. Air Canada, 2024 BCCRT 149).
- Air Canada argued, in effect, that the chatbot was a separate legal entity. The tribunal called that “a remarkable submission.”
- The tribunal also said customers should not have to double-check one part of a company’s website against another.
- The award was small. The reasoning applies to any business that lets a bot answer questions about money.
On November 11, 2022, Jake Moffatt’s grandmother died in Ontario. The same day, Moffatt went to Air Canada’s website to book a flight from Vancouver to Toronto and asked its chatbot about bereavement fares. The bot said a traveller who had already flown could submit the ticket for a reduced bereavement rate within 90 days of the ticket being issued. Moffatt booked two flights, for $794.98 and $845.38, and applied on November 17.
Air Canada’s own bereavement page said the opposite: the policy did not apply after travel. On February 8, 2023, an Air Canada representative admitted the chatbot had used “misleading words” and said the company would update it. It still refused the refund. Moffatt took it to British Columbia’s Civil Resolution Tribunal.
- November 11, 2022Moffatt asks the chatbot about bereavement fares
- November 17Moffatt applies for the reduced rate
- February 8, 2023An Air Canada representative admits “misleading words”
- February 14, 2024Tribunal finds Air Canada liable and orders $812.02
What happened in Moffatt v. Air Canada?
On February 14, 2024, Tribunal Member Christopher C. Rivers found Air Canada liable for negligent misrepresentation by its chatbot. He ordered the airline to pay $812.02: $650.88 in damages, the difference between what Moffatt paid and the bereavement fare, plus $36.14 in interest and $125 in tribunal fees. He refused to deduct a $200 coupon Air Canada had offered.
While a chatbot has an interactive component, it is still just a part of Air Canada’s website. It should be obvious to Air Canada that it is responsible for all the information on its website. It makes no difference whether the information comes from a static page or a chatbot.
Tribunal Member Christopher C. Rivers, Moffatt v. Air Canada, 2024 BCCRT 149, paragraph 27
Is a business liable for what its chatbot tells customers?
In this decision, yes, on ordinary negligence reasoning. The tribunal found Air Canada owed its customer a duty of care, the chatbot’s statement was inaccurate, the airline did not take reasonable care to keep it accurate, and the customer reasonably relied on it and lost money. The bot being automated was no defence.
Two parts of the reasoning matter for any business. Air Canada argued that the correct policy was on another page. The tribunal said the airline did not explain why that page was more trustworthy than its chatbot, or why customers should have to check one part of its website against another. It also found the customer had no way to know which part of the site was accurate.
Keep the limits in view. This is a small claims decision from British Columbia, and it binds nobody in the United States. What it shows is how an ordinary judge looks at the argument that the bot said it, not us. That argument did not survive one paragraph.
When does a chatbot hurt more than it helps?
When the question has money or an exception attached and the bot answers from memory instead of from one maintained source. Refunds, cancellations, special fares, price for a specific case and anything emotional all fit that description. Moffatt’s question hit two of them at once: a refund rule, asked on the day someone in the family died.
| Question type | Risk if the bot is wrong | What to do |
|---|---|---|
| Hours, address, parking | Low, if pulled from one maintained source | Let the bot answer |
| Order or appointment status | Low to medium, if read live from your system | Answer from the system, never from memory |
| Refunds, cancellations, exceptions | High: policy with money attached | Quote the policy page word for word and link it, or hand off |
| Price for a specific situation | High | Quote the published price list or hand to a person |
| Anything emotional: a death, an illness, a complaint | High | Hand to a person quickly |
| Legal, medical or financial advice | Highest | Do not answer. Hand off |
How do you stop a chatbot from promising what you can’t deliver?
Give it one source of truth per policy, let it answer only from that source, and list the topics it must hand to a person. Then test it with real customer questions before launch, read a sample of conversations every week, and update it the same day a policy changes. Name one person who owns it.
- Pick the policies the bot may talk about, and point each one at a single page you maintain.
- Write a hand-off list: refunds outside the standard rule, complaints, anything about a death or illness, and anything that sounds like advice.
- Collect 30 questions customers actually asked, including the awkward ones, and run them before launch.
- Read a sample of conversations every week. Wrong answers show up there long before a complaint does.
- When a policy changes, change the bot the same day. Put one person’s name on that job.
- If the bot tells a customer something wrong, decide in advance to honor it. The tribunal made Air Canada pay anyway, more than a year later.
The discipline is the same one behind any agent: written instructions it can follow, and a clear line for when a person takes over. Both have their own walk-throughs, on writing an SOP an agent can follow and on where approval belongs.
Should a small business use a customer service chatbot at all?
Yes, if you can name the questions it will answer and keep the answers current. A bot that handles hours, directions, booking and order status well takes real load off the front desk. A bot turned loose on every question, with nobody reading its conversations, is a liability you are paying for.
If your policies live in people’s heads, write them down before you buy a bot, because the bot will make them up. If nobody on the team has time to read a sample of conversations each week, wait. And if an AI assistant elsewhere is already describing your business wrongly, that is a separate problem with its own fix.
Questions people ask
Does Moffatt v. Air Canada apply in the United States?
Not as precedent. It is a small claims decision from British Columbia’s Civil Resolution Tribunal and binds no US court. Its reasoning is ordinary negligent misrepresentation: a company is responsible for the information on its website, whether it comes from a static page or a chatbot. Ask your own counsel how your state treats it.
How much did Air Canada have to pay?
A total of $812.02: $650.88 in damages, $36.14 in pre-judgment interest and $125 in tribunal fees. Moffatt had claimed $880. The tribunal refused to offset a $200 coupon Air Canada had offered, because Moffatt never accepted it.
Does a disclaimer on the chatbot protect the business?
The Air Canada decision did not turn on a disclaimer, so it does not answer that. It did reject the argument that the correct answer on another page excused the wrong one. A disclaimer does not replace keeping the bot’s answers accurate.
What questions should a customer service chatbot never answer?
Anything legal, medical or financial that reads as advice, exceptions to refund or cancellation rules, prices for a specific case that are not on your published list, and conversations about a death, illness or complaint. Those go to a person, quickly, with the conversation attached.
Is an AI agent riskier than a chatbot here?
Usually, because an agent can act as well as answer: issue a refund, change a booking, send an email. The same rule applies to what it says, and the damage from a wrong action is bigger. Put a person on any action that moves money or changes a customer’s terms.
Sources
- Civil Resolution Tribunal of British Columbia, Moffatt v. Air Canada, 2024 BCCRT 149 (decision issued February 14, 2024): Every fact and quotation about the case: the chatbot’s answer, the fares paid, the “misleading words” admission, the paragraph 27 reasoning, the finding on double-checking the website, and the $812.02 order.
Published . Last reviewed .
Want this in your business?
Book a call and we will scope exactly which part of your operation an agent should take first.
Book a call with the team