Clifford Chance

The EU Introduces New Rules on AI Liability

30 January 2025

In a representative survey of 2020, the unclear regulation of liability ranked amongst the top three barriers to the use of artificial intelligence (the "AI") by European companies. Now the EU is aiming to reverse this trend and set clear rules, facilitating the process of determining who is liable for the developing and implementing AI products.

Existing national liability rules, particularly those based on fault, are ill-equipped to address liability claims arising from damage caused by AI-enabled products and services. These rules often require victims to demonstrate a wrongful act or omission by the individual responsible for the damage. This also applies to the Czech legal framework with reference to the provisions of Section 2909 et seq. of Act 89/2012 Coll., the Civil Code, which would currently be used to assess liability for AI. However, the unique attributes of AI, such as its complexity, autonomy, and its opacity, often referred to as the "black box" effect (meaning a lack of clear understanding or transparency regarding its internal processes and decision-making), can make it challenging or very costly for victims to identify the responsible party and meet the criteria for a successful liability claim. Specifically, when seeking compensation, victims and deployers might face substantial initial costs, legal uncertainty and significantly prolonged legal proceedings compared to cases not involving AI.

Download PDF