Who's liable when your AI gets it wrong?
2026 has brought a wave of lawsuits and fines over AI that misled people. The pattern is clear: when your AI gets it wrong, the business is the one that pays.
A run of cases through 2026 has settled a question many businesses hoped to avoid: when an AI tool gets something wrong, who is responsible? The answer, increasingly, is the business that deployed it.
In May, Pennsylvania sued Character.AI to stop its chatbots posing as licensed professionals, after bots presented themselves as psychiatrists and handed out medical advice, one reportedly offering a fake medical licence number (NPR). In the courts, hallucinations kept costing lawyers real money: in one Oregon case, two attorneys were fined US$110,000 after an AI tool led them to cite authorities that did not exist.
These build on a precedent set back in 2024, when a tribunal held Air Canada responsible for a bereavement discount its own chatbot had invented, ordering the airline to honour the made-up policy. The principle has only hardened since: you cannot blame the bot.
"The AI did it" is not a defence
The thread running through every one of these cases is the same. A business put an AI tool in front of customers, or relied on it for real work, the tool said something confident and wrong, and the business, not the vendor and certainly not the AI, carried the consequences.
That is worth taking on board before you point an AI at anything that matters. An AI agent talking to your customers is talking for your business, with all the liability that carries. An AI answer used in your work becomes your answer the moment you act on it.
The cheap insurance
None of this is a reason to avoid AI. It is a reason to put a few unglamorous controls around it: decide where a human has to check before AI output reaches a customer or a decision, be clear about what your AI is and isn't allowed to say, and keep a record of how it is set up. The businesses in the headlines skipped exactly these steps (the AI risks most businesses aren't pricing in).
The technology is new. The principle is old: you are responsible for what your business tells people, whoever, or whatever, does the telling.
Want to know where your team actually stands?