Who is liable when AI makes a mistake?

A machine is not a legal person. It cannot be sued or punished. Responsibility is therefore spread across the people and organizations in the chain. If a doctor adopts an AI recommendation without checking it, she remains responsible. If a product with built-in AI is faulty, the manufacturer can be held liable. Courts have already fined lawyers who cited court rulings invented by AI.

With the AI Act, the EU scales obligations according to risk: stricter rules apply to AI used in job applications, loan approval, or medicine than to a customer service chatbot. Switzerland is taking its own path. In 2025, the Federal Council decided to implement the Council of Europe's AI convention and to adapt existing laws in a targeted way, rather than create a separate AI law.

For companies, this means: whoever uses AI needs clear rules on who checks results and who approves them.