Air Canada's Chatbot Liability
Featuring Jake Moffatt
In late 2022 a grieving traveler asked Air Canada's website chatbot about bereavement fares, and the bot confidently told him something that simply was not true. When he tried to claim what he'd been promised, the airline refused, then mounted a legal defense so audacious it made headlines: it argued the chatbot was a separate entity responsible for its own words. A Canadian tribunal ruled in early 2024, and the dollars at stake were almost beside the point.
For founders and operators shipping customer-facing AI, this case forces a question you probably haven't answered: if your bot confidently says something wrong tomorrow and a customer acts on it, who owns that? It sharpens the decision of how to govern AI you've deployed at scale. The precedent the tribunal set, and what it means for everyone building on top of these tools, is the part you'll want to read.
Frequently asked questions
What happened with the Air Canada chatbot lawsuit?
Air Canada's website chatbot gave a traveler wrong information about bereavement fares in late 2022, and when he tried to claim what he had been promised, the airline refused. A Canadian tribunal ruled against Air Canada in early 2024, rejecting its defense and holding the company responsible for what its bot said. The case became a landmark on who is accountable for AI-generated answers.
What was Air Canada's defense in the chatbot case?
Air Canada argued that its chatbot was a separate legal entity responsible for its own words, so the airline should not be liable for the bot's promise. The tribunal flatly rejected this, treating the chatbot as part of Air Canada's website and the airline as responsible for all information on it. The argument made headlines for how unusual it was.
Why does the Air Canada chatbot ruling matter for AI governance?
It set an early legal precedent that a company owns what its customer-facing AI tells people, even when the AI is wrong. For any business deploying chatbots at scale, the ruling closes off the idea that you can disclaim responsibility for the model's mistakes. It reframed AI deployment as an accountability and governance problem, not just a tech one.
What can founders learn from the Air Canada chatbot case?
If you ship customer-facing AI, you own its mistakes, so you need governance over what it can promise before it goes live, not after a customer acts on a bad answer. The lesson is to decide in advance how you constrain, monitor, and stand behind AI you deploy. CaseBook turns this into a move you apply to your own company, with an AI coach that reads your answer.