When a chatbot invented
a refund policy
Air Canada's support bot told a grieving customer he could book a full-price flight and retroactively claim a bereavement discount. The policy didn't exist. The airline said the bot was "a separate legal entity." The tribunal disagreed.
The chatbot promised a policy that didn't exist
In November 2022, Jake Moffatt's grandmother died. He needed to fly to Toronto for the funeral. He visited Air Canada's website and asked the chatbot about bereavement fares.
The bot told him he could book a regular flight and then submit a refund request within 90 days to receive the bereavement rate retroactively. So he booked a $1,206.84 flight, attended the funeral, and filed the claim.
Air Canada denied it. The bereavement-fare policy, they said, required booking at the bereavement rate before travel — not retroactively. The chatbot had invented a process that didn't exist.
When Moffatt took them to a civil tribunal, Air Canada's lawyers argued the chatbot was "a separate legal entity, responsible for its own actions." The tribunal member called that "remarkable" and ruled in Moffatt's favor.
What the chatbot actually said
What it cost
"While a chatbot can be a useful tool, Air Canada has failed to explain why its chatbot was not under its control. It should be apparent to Air Canada that it is responsible for all the information on its website."
— Tribunal Member Christopher Rivers, Civil Resolution Tribunal of British Columbia
Three review criteria that would have caught this
Each criterion below maps to a real review task you can configure in the sample builder. A certified reviewer checks every chatbot response against these before it reaches a customer.
Verify all policy claims against source documents
Every factual claim about a policy, procedure, or benefit must cite the exact section of the source document. If the AI invents a process, the reviewer flags it immediately.
No contradiction of published policy
The AI's response is compared against the actual published policy. If the bot says "submit within 90 days" but the policy says "book at bereavement rate before travel," the reviewer flags the contradiction.
No invented legal or financial obligations
If the AI creates a commitment the company is bound to honor — a refund, a policy exception, a price guarantee — the reviewer checks whether the company actually offers it. Invented obligations are the highest-severity failure.
Paste any chatbot response. See what gets flagged.
This is a simplified version of what our reviewers see. Paste a customer-support response (yours or a competitor's) and run the check. The criteria above are applied automatically.
Don't wait for the tribunal
Every chatbot response is a liability. Put certified reviewers between your AI and your customers. 50% off your first $10 — live in under 5 minutes.