Bitkom on the EU AI Liability Directive  

Today, the European Commission publishes its proposal on the AI Liability Directive, which regulates liability in the use of artificial intelligence. Susanne Dehmel, Member of the Bitkom Executive Board, states:

"We welcome the fact that the draft AI Liability Directive addresses fundamental questions on liability in the event of damage when using AI. Artificial intelligence is an extremely important future technology for the German economy. The vast majority of companies (65 percent) see the use of AI as an opportunity for their own business - but only nine percent use AI so far. In order to advance the use of AI in Germany, legal clarity for its practical application must be created as quickly as possible.

By referring to the relevant definitions from the AI Act, the AI Liability Directive ensures consistency and avoids legal uncertainties. Furthermore, it is important for Europe that the directive does not generally classify AI systems as a source of danger to life and health. Manufacturers or operators of an AI system should not be subject to strict liability regardless of fault. This gives manufacturers or operators of AI systems the opportunity to limit their liability risk by acting responsibly. We welcome the fact that in the event of a court-ordered disclosure of data or processes, the legitimate interests of the manufacturers, such as the protection of business secrets, are also taken into account.

However, we are more critical of the reversal of the burden of proof provided for in the proposal: If it cannot be clearly established whether the damage caused is causally related to a breach of duty on the part of the manufacturer or operator, the latter will first be presumed to be at fault - this presumption must then be rebutted by the manufacturer or operator himself by proving 'other plausible explanations'. So far, the proposed directive leaves open what this concretely means. We need more clarity on this point.”