Your AI's Words Are Your Words
A Canadian tribunal set the floor: an AI cannot be answerable. Accountability snaps back to whoever deployed it.
THE GIST
- Air Canada argued its chatbot was “a separate legal entity.” The tribunal's answer was permanent: your AI's words are your words.
- Trust is an architecture: explainability + accountability + recourse. AI can carry only the first.
- Only 46% of people trust AI while 66% use it — and the EU AI Act's high-risk rules bind in August 2026.
Jake Moffatt was booking flights to his grandmother's funeral when Air Canada's chatbot told him he could apply for the bereavement fare after the trip. The bot was wrong — it contradicted the very policy page it linked to. When Moffatt asked for the refund, Air Canada refused. And in the tribunal, the airline made an argument that belongs in business-school case studies forever: the chatbot, it said, was "a separate legal entity that is responsible for its own actions."
The tribunal's answer was short and permanent: no. A company is responsible for all the information on its website, chatbot included. The damages were CAD $812 — trivial. The principle was not: an AI cannot be answerable. Accountability snaps back to whoever deployed it.
This is why trust, in an AI-saturated organization, stops being an atmosphere and becomes an architecture — with three load-bearing parts:
Reliance is not earned by being right more often. It is earned by being answerable when wrong.
- Explainability: we can show why.
- Accountability: a named human answers for it.
- Recourse: the person affected can challenge it and reach a human.
Notice what AI can carry: exactly one of the three. It can generate an explanation. It cannot be held responsible — responsibility requires someone who can answer for the call. And it cannot offer recourse — recourse requires someone with authority to make it right. Automation can log every step but cannot answer for one. Only a person can stand behind a decision.
The market knows something is missing here. KPMG's global study found just 46 percent of people are willing to trust AI — while 66 percent already use it. People are relying on systems they don't trust, which is not a stable arrangement; it's a withdrawal from the trust account on every transaction. And the regulators have stopped waiting: the EU AI Act's high-risk obligations — covering AI in employment and credit decisions — become binding in August 2026. The accountability regime is no longer a forecast.
TAKE IT TO THE FLOOR
For any system that makes or shapes decisions about people, ask: when this system is wrong about someone, what happens next — and who answers? If the answer is a name, you're building Trust Capital. If the answer is "the model said so," remember the airline.