Who is responsible when an AI makes a mistake?

When an AI makes a mistake, people instinctively look for someone responsible on the side of the machine or its supplier. For your customers, your employees or your partners, the answer is simpler: it was your company that took the decision, sent the letter or applied the price. It is first and foremost your company that has to answer for it.

It may sometimes be able to turn to its supplier, depending on the contract and the nature of the defect. But above all, it will have to show that it used the tool responsibly: that it knew what it was for, who decided what, which checks were planned and that they were carried out. That case is prepared before the incident, not after.

• Suppliers’ contracts often limit their liability.The general terms of AI offers protect the supplier first.
• Decisions taken with AI leave few traces.Afterwards, it is rarely known what the tool proposed and what the person approved.
• The chain of decision is often blurred.Between the business, IT and the supplier, each thinks the other has checked.

1. 01For each use, write down who decides, who checks and who answers in case of error.
2. 02When the stakes justify it, keep a record of what the AI proposed and what was approved.
3. 03Have a lawyer review the liability clauses in your contracts with AI suppliers.

We help make visible who decides what, so that you can show it the day you are asked.

• What do we do when AI gets it wrong?→
• How far should AI be allowed to decide on its own?→
• What does the EU AI Act change for our company?→
• Who should govern AI in the company?→