Air Canada’s chatbot

In November 2022, after the death of their grandmother, Jake Moffatt used Air Canada’s website to book flights. While doing so, Mr Moffatt asked the airline’s chatbot about bereavement fares. The chatbot told him that a discount was available for people travelling because of a death in the family and indicated that a reduced fare could be requested for the reduced fare within ninety days of the ticket being issued.²

That was wrong. Air Canada’s own bereavement policy page stated that the policy “does not apply to requests for bereavement consideration after travel has been completed.”¹ When Mr Moffatt applied for the reduced fare, the airline refused, and Mr Moffatt took the claim to British Columbia’s Civil Resolution Tribunal.

The airline’s defence

Air Canada argued that it could not be held liable for information provided by one of its agents, servants or representatives, including the chatbot. Summarising the airline’s position, tribunal member Christopher C. Rivers wrote: “In effect, Air Canada suggests the chatbot is a separate legal entity that is responsible for its own actions.”¹

The tribunal described the chatbot as “just a part of Air Canada’s website” and rejected the argument plainly: “It should be obvious to Air Canada that it is responsible for all the information on its website.”¹ It also rejected the idea that Mr Moffatt should have checked the chatbot’s answer against the airline’s policy pages elsewhere on the site, finding that a consumer cannot be expected to double-check information found on one part of a website against another.²

The ruling

The tribunal found that Mr Moffatt had made out a claim of negligent misrepresentation. Mr Moffatt had paid $1,630.36 for the flights, against an estimated $979.48 at the bereavement fare. The tribunal ordered Air Canada to pay $812.02: $650.88 in damages for the difference, $36.14 in pre-judgment interest and $125 in tribunal fees.¹

The sum was small. The principle was not, and the decision has been widely cited since by lawyers and commentators discussing who is accountable for AI-generated advice.

Why it matters

Customers do not distinguish between channels. To a customer, the chatbot is the company. Its answers carry the same authority as a web page, a letter or a call-centre agent.

Generative AI will sometimes be confidently wrong. Any system that answers open questions in natural language can produce answers that sound right and are not. Organisations need to plan for that, not hope it will not happen.

Arguing that the AI is responsible is a reputational own goal. Air Canada’s defence turned a minor customer dispute into an internationally reported story about a company disowning its own technology.

Lessons for leaders

1. Own every answer your AI gives. Treat chatbot responses as statements made on behalf of the organisation, with the same legal and reputational weight.
2. Restrict AI on high-stakes topics. For prices, refunds, eligibility and legal rights, ground answers in approved policy text, or hand the customer to a person.
3. Plan how you will put things right. When the AI gets it wrong, the quickest and cheapest response is usually to honour what the customer was told and fix the system.

Questions to ask in your organisation

1. Which of our customer-facing AI tools give information about prices, entitlements, refunds or legal rights?
2. If one of them gave a customer wrong information today, what is our policy for putting it right, and who decides?
3. Would our legal team ever argue that the AI, not the organisation, was responsible, and are we comfortable with how that would look?