In principle, the general liability rules also apply to the use of AI-based products. If the architect can be proven to be at fault or negligent, he or she will continue to be liable. However, the use of AI-based products entails some special features compared to other products. In particular, it is not so easy to understand how AI applications arrive at their results. This phenomenon is known as the “black box effect”.
From a liability law perspective, this is problematic as it is difficult to prove if damage and therefore potential liability claims have arisen due to faulty AI. In this mixed situation, there is a risk that the architect, as the user, will become solely liable. To prevent this, the EUEU Europäische Union has amended the Product Liability Directive and adapted the rules of evidence for the use of AI software. Due to the new regulations, in particular the simplification of the burden of proof and the duty of disclosure, it is much easier to hold the manufacturer of the AI software liable in the event of damage.
It can therefore be roughly stated:
- If the fault is due to the AI, the manufacturer is liable. With the help of the new rules of the Product Liability Directive, this should also be easier to enforce in terms of evidence.
- However, if the error is due to the fault of the architect, liability applies. What this will mean in concrete terms in connection with the use of AI remains to be seen and will probably also be decided by the courts.
