AI Solutions

    Air Canada's Chatbot Lawsuit: What Every Business Must Learn About Chatbot Liability

    A court held Air Canada responsible for what its chatbot said. Learn what happened, why it matters, and how to reduce your own chatbot liability risk.

    8SolveAI Team4 min read

    In 2024, a Canadian tribunal ruled that Air Canada was responsible for wrong information its own chatbot gave to a customer. The amount was small, but the principle was not: a company cannot treat its chatbot as a separate entity and walk away from what it says.

    If your business uses, or is about to use, an AI chatbot, this case is worth understanding. It shows where the real risk sits and what you can do about it.

    What happened

    A passenger, Jake Moffatt, used Air Canada's website chatbot to ask about bereavement fares after a family member died. The chatbot told him he could buy a full-price ticket and apply for a bereavement discount afterwards, within a set period. That contradicted the airline's actual policy, which did not allow retroactive bereavement refunds.

    He followed the chatbot's advice, then applied for the refund. The airline refused, pointing to its policy page. The dispute went to British Columbia's Civil Resolution Tribunal in a case known as Moffatt v. Air Canada (2024 BCCRT 149).

    What the tribunal decided

    The tribunal found Air Canada liable for negligent misrepresentation. Several points stood out, as summarised by law firms such as Mondaq's analysis and CTV News:

    • The chatbot is part of your website. Air Canada argued, in effect, that the chatbot was a separate entity responsible for its own actions. The tribunal rejected that. The company is responsible for all the information on its website, including what the chatbot says.
    • A correct policy page elsewhere did not save the airline. The customer had no reason to believe the chatbot was less reliable than another page on the same site.
    • Reasonable care applies. The tribunal found Air Canada had not taken reasonable care to make sure its chatbot was accurate.

    The compensation was modest. The importance of the case is as a precedent and a warning.

    What this means for your business

    This was a decision from one tribunal in one country, and it is not a universal rule. But the reasoning is simple enough that it is likely to be persuasive elsewhere, and it matches how most people think: if your chatbot says it, you said it.

    That changes how you should think about chatbots:

    1. Accuracy is a business and legal issue, not only a product-quality issue.
    2. You cannot outsource responsibility to the technology or the vendor. A vendor may share some responsibility in a contract, but your customers will deal with you.
    3. Documentation matters. If something goes wrong, records of how you tested and monitored the chatbot support a reasonable-care argument.

    This article is general information, not legal advice. Talk to a qualified lawyer about your situation.

    How to reduce your chatbot liability risk

    1. Ground answers in approved content

    A chatbot that answers from your approved policies and documents is far less likely to invent terms. Retrieval-based designs, which we explain in our RAG chatbot guide, help a lot. We also cover practical controls in our article on stopping chatbot hallucinations.

    2. Narrow the scope

    Tell the bot what it must not discuss, and let it say "I'm not sure, let me connect you with a person" for anything outside its knowledge. Our guide to chatbot prompt design shows how.

    3. Treat high-risk topics carefully

    Refunds, pricing, contracts, legal, medical, and financial questions deserve extra care. Options include a fixed approved answer, a link to the official page, or a handover to a person.

    4. Be transparent

    Make it clear that the customer is talking to an automated assistant, and that important information should be confirmed with official sources. Experts have noted that a clear disclaimer might have helped, though a disclaimer is not a substitute for accuracy.

    5. Test and keep records

    Test with real and tricky questions before launch, and keep the results. Our chatbot testing checklist lists what to cover. Keep logs of conversations so you can review and correct problems.

    6. Monitor after launch

    Review real conversations regularly, track unanswered and escalated questions, and fix wrong answers quickly. Our guide to chatbot KPIs shows what to measure.

    7. Review your contracts and terms

    Check your vendor agreements and your website terms with a lawyer. Some companies add terms about reliance on automated responses, though these may not protect you in every situation.

    8. Make human help easy

    If a customer is unsure, a person should be one click away.

    The bigger lesson

    The Air Canada case did not show that chatbots are too risky to use. It showed that a chatbot is part of your business, and should be managed like any other customer-facing channel: trained, tested, supervised, and corrected when it makes mistakes.

    Companies that do this get the benefits of faster, always-on support and avoid the nasty surprises.

    Build a chatbot you can trust

    We build AI chatbots grounded in your approved content, with clear scope rules, testing, and human handover. See our chatbot development service, or book a free consultation.

    Have a project in mind?

    Tell us what you want to build and we will suggest the simplest path to get there.

    Keep reading

    All articles