On November 11, 2022 — Remembrance Day — Jake Moffatt’s grandmother died. He did what people do: found the first flight he could from Vancouver to Toronto, on Air Canada, and asked the airline’s support chatbot about bereavement fares. The chatbot was reassuring. He could book at full price, it said, and apply for the reduced fare within ninety days of ticketing — a refund, after the fact. So he flew east, attended the funeral, and submitted his application. Air Canada refused. Its actual policy, published elsewhere on the same website, required the discount to be requested and approved before travel. The chatbot had been describing a policy that did not exist.
A wrong answer from customer service is not news; people have been misquoted by helplines since there were helplines. What happened next is the story. Moffatt filed a claim with British Columbia’s Civil Resolution Tribunal, the province’s online small-claims forum, and Air Canada — over a sum that would eventually be fixed at $812.02, refund, interest and tribunal fees included — offered a genuinely ambitious defence. It could not be held liable, the airline argued, for information provided by its chatbot, because the chatbot was, in effect, a separate legal entity, responsible for its own actions.
Sit with that defence for a moment, because no airline would attempt it with a human. If a reservations agent had misquoted the bereavement policy, Air Canada would not have told the tribunal the agent was a separate legal entity. The employee wears the company on their lanyard; their words are the company’s words, which is exactly why companies train them, script them and stand behind them. The argument becomes sayable only when the speaker is software — only then can a business put its logo on a mouth and insist, the moment the mouth becomes expensive, that nobody was speaking. Air Canada wanted the bot’s answers authoritative enough to close a sale and deniable enough to void one.
Tribunal member Christopher Rivers was unmoved. A chatbot, he wrote in February 2024, is “still just a part of Air Canada’s website,” and “it makes no difference whether the information comes from a static page or a chatbot.” He found negligent misrepresentation, and noted that the airline never explained why customers should trust one part of its website over another. The decision binds no one beyond the two parties; it is a small-claims ruling in one province, not a new law of the internet. That smallness is part of the point. It took no theory of artificial intelligence to resolve the case, because the principle underneath it predates software entirely: whoever speaks in your name speaks for you.
It is worth adding what the ruling does not say. Nothing in it establishes that the chatbot was a large language model, or any kind of generative system; “hallucination” was the word the headlines reached for, but the tribunal needed nothing so modern. Whatever the bot was doing when it produced those ninety days, the finding turned only on whose website it sat on. If there is a hallucination anywhere in this record, it is not in the chat window. It is the legal theory — the belief, hardly unique to one airline, that you can put words in a machine’s mouth and then disown the machine.