Air Canada's Chatbot Ruling: You Are Liable for What Your AI Says

What happened

After his grandmother died, Jake Moffatt asked Air Canada's chatbot about bereavement fares. It told him he could buy a full-price ticket and apply for the reduced fare afterwards. That was not the airline's policy, and his claim was refused.

Air Canada argued that the chatbot was responsible for its own statements. The British Columbia Civil Resolution Tribunal rejected that, finding the airline responsible for all the information on its website, whether it came from a static page or a chatbot. It found Air Canada liable for negligent misrepresentation.

Why this matters beyond airlines

Customers have no way of knowing which parts of your website are written by people and which are generated by AI, and the tribunal saw no reason they should. If your AI tells a customer something, you said it.

The risk is greatest where answers carry money, rights or safety: refunds, eligibility, pricing, medical guidance, legal or financial information.

What reliability testing looks like

• Test accuracy against the real questions your customers ask, not a curated demo set.
• Ask the same question many times and in different words, and check the answers stay consistent.
• Check that answers are grounded in approved sources such as your current policies.
• Check that the AI says it does not know rather than inventing an answer.
• Re-test after every model update, because behaviour can drift without any change on your side.

The takeaway

A chatbot can be right most of the time and still create liability the one time it is wrong. The only way to know how often it is wrong, and where, is to test it systematically and keep testing it.