Air Canada's chatbot ruling: when a screenshot becomes the evidence

In February 2024, a tribunal in British Columbia gave Air Canada a lesson in evidence.

A customer, Jake Moffatt, had asked the airline's chatbot about bereavement fares and got an answer that turned out to be wrong. When the airline refused to honour the discount, he took them to the Civil Resolution Tribunal. The tribunal ordered the airline to compensate him ^[1].

The airline's defence was unusual. It argued it couldn't be held liable for information from its "agents", including a chatbot. In effect, the tribunal said, the airline was suggesting the chatbot was "a separate legal entity that is responsible for its own actions". The tribunal didn't accept that. Air Canada is responsible for everything on its website, whether it comes from a static page or a chatbot ^[1].

The evidence is the interesting part. The customer had a screenshot of the chatbot's answer. An airline representative had already admitted that the words were misleading. On the fare itself, the airline offered no evidence to the contrary, so the tribunal assumed such evidence wouldn't have helped it ^[1].

In other words, the customer kept a record. The airline didn't.

Like a receipt you keep yourself

Think of it like this. You pay for something, and the shop gives you a receipt. A month later there's a dispute, and the receipt is the proof. But what if the shop doesn't give you a receipt, and you didn't keep one yourself? Then it's your word against theirs.

That's roughly what happened here. The chatbot said something, the customer kept a screenshot, and when the airline couldn't show what the chatbot actually said, the screenshot became the evidence.

It's a simple point, but one that deserves attention. When an AI talks to your customers, who keeps the record?

What the AI Act says

Since 2 August 2026, Article 50 of the AI Act has required that people be told when they are interacting with an AI system, unless it's obvious from the context ^[2]. There are exceptions, for example for AI that detects or prevents fraud, but the rule is disclosure.

The idea is transparency: people should know when they're talking to a machine.

What Article 50 doesn't say is who keeps the record of the conversation, or how. That's a separate question, and the Air Canada case shows why it matters.

The last-mile problem

Most organisations already keep logs of customer interactions. The question is whether those logs hold up when someone questions them.

A log your own team can edit is easy to challenge, especially if the other side turns up with a screenshot. Access controls and monitoring are the foundation, but they don't solve the last-mile problem: when the people who keep the logs could also change them, how do you show that nobody did?

What a digital witness does here

Here, the thing to seal is each answer the chatbot gives, or each conversation when it ends, together with the notice that the customer was talking to an AI.

You send a fingerprint of the record to WitniumChain through the REST API, the SDK or the MCP server, and get a witness ID back. The record itself stays with you. To check it later, anyone you share it with can calculate the fingerprint again and compare it with the sealed one. Change one word and it won't match.

Who can check it? Customer service, the customer, an ombudsman or a court.

What it shows is that the record of what the chatbot said, and when, hasn't changed since it was sealed. What it doesn't show is that the answer was correct, or settle whether the company is bound by it. Those are separate questions.

The first 2,000 witnesses a month are free. Read more in the documentation.

How it would work

1. The chatbot sends an answer. Your system seals a record of the question, the answer and the AI disclosure.
2. The customer later disputes what the chatbot said. You share the sealed record.
3. They check the fingerprint. If it matches, the record is unchanged. If it doesn't, something has been edited.

What this means for you

If you run a customer-facing chatbot, here are three questions worth asking now:

• Do we keep a record of what the chatbot says, conversation by conversation?
• If a customer turned up with a screenshot six months from now, could we show our own version of what was said?
• Who in the organisation could edit those records, and would a change show up?

The Air Canada case didn't turn on new technology. It turned on a basic principle: the side without a record starts on the back foot.

What to watch

Article 50 of the AI Act applies from 2 August 2026 ^[2]. The transparency duty is clear. The question of records is one each organisation has to answer for itself, and the answer shapes what happens on the day someone questions a chatbot's promise.