Chatbot Misrepresentation Liability
Tags: Case Studies, Tools & Platforms
TL;DR
- A Canadian court held Air Canada liable for a discount promised by its AI chatbot.
- Rulings signal brands can be responsible for chatbot outputs, not just website fine print.
Why it matters for HK marketers: If your chatbot misleads customers, you may eat the cost — governance must cover prompts, guardrails, and monitoring.
What happened
- AI agent promise: The chatbot offered a discount; the court enforced the representation against the airline.
- Liability principle: Brands are accountable for claims made by automated agents interfacing with consumers.
Risk controls for chatbots
- Content governance: Pre‑approved prompts, restricted intents, and escalation to human agents.
- Monitoring: Log reviews for hallucinations and “learned nastiness”; rapid rollback procedures.
- Data handling: Define retention, training use, and deletion pathways for user inputs.
So what for marketers
Tighten chatbot guardrails, disclaimers, and human handoffs; continuously test for pricing/promo claims that could bind your company.
← Back to Knowledge Base