While the EU’s AI Act has entered into force in 2024, this is not (yet) the case for any binding legislation on liability for AI-related damage at the EU level. The EU institutions issued a proposal for a Directive on AI liability (AILD) in September 2022, but withdrew it in February 2025. As opposed to what the name of the Directive suggested, it did not harmonise the substance of liability. These issues were and still are left to be regulated by the Member States. As the regulatory landscape on AI in general is still evolving and discussion on AI liability will continue, this contribution presents some takeaways from the failed Directive. In addition, this chapter delves into traditional concepts of civil liability and their compatibility with liability for damage caused by artificial intelligence and shows examples of national laws of Member States. It will further explore selected national civil liability laws and approaches as to their suitability to tackle damage caused by artificial intelligence. The AI Liability Directive strived to mitigate the information asymmetry between a wrongdoer and an injured party where AI is involved to a certain extent. It did not present any solutions to the often difficult question of how to attribute the behaviour of AI systems to a human based on the prevailing human-centric approach to AI.