A grieving man asked Air Canada's chatbot about discounted fares for a death in the family. The bot answered confidently — and got it completely wrong. When he sued, the airline tried a remarkable dodge: it argued the chatbot was its own separate thing, not really the company's responsibility. A tribunal rejected that and made them pay.
That excuse — 'blame the bot, not us' — turns out to be the pattern, not the exception. A study of 1,406 real AI incidents found nearly half came from ordinary software like chatbots and recommendation tools, not sci-fi robots. Plain customer-service bots, handed too much authority and left unsupervised, confidently telling people things that simply aren't true.
So how does it touch you? Next time a company's chatbot quotes you a price, a policy, or a refund rule, it may be flatly wrong — and the company's first instinct will be to say the bot did it, not them. This case is the early proof that 'the computer said so' isn't a free pass. You can still hold them responsible.