On September 25th, 2020, the European Parliament Research Service (EPRS) released a new European added value assessment (EAVA) on civil liability regime for artificial intelligence.
According to the study, ‘’the revision of the EU civil liability regime for artificial intelligence systems (AI) would likely generate substantial economic and social added value’’:
“by 2030, EU action on liability could generate €54.8 billion in added value for the EU economy by stepping up the level of research and development in AI and in the range of €498.3 billion if other broader impacts’’.
Source: EAVA on Civil Liability Regime for Artificial Intelligence
However, to generate these substantial added value, ‘’a clear and coherent EU civil liability regime for AI’’ is necessary, as a comparative legal analysis made by the EPRS ‘’indicates great divergence between Member States in terms of their current rules and their degree of flexibility to adjust to the new challenges related to AI’’:
“This comparative analysis indicates that in the absence of common EU action, it is very likely that very divergent practices and interpretations might emerge in the Member States, in some situations potentially introducing obstacles to the functioning of the internal market. (…) Common EU action on AI liability is necessary as it would bring added value that could not be achieved by the individual actions of Member States’’
The findings of the study are divided into 5 main sections:
- 2. AI liability — Why regulate? Socio-economic functions of civil liability rules in the AI context
- 3. Liability — how is it regulated? Current EU law liability framework and its application to AI
- 4. Liability — how is it regulated? Regulation of liability in national law and its possible application to AI
- 5. AI liability – How to regulate in the future? Policy options and approaches
- 6. Europe an added value — Quantitive assessment
The EAVA focuses on the main points of discussion related to liability regime for AI systems. The findings of the study explores the legal issues of AI liability at the light of the EU Product Liability Directive (PLD), or focusing on the existing or currently drafting regimes establishing strict liability or open/mixed/ narrow systems of liability for AI systems. The study offers key findings and interesting policy comparison as a harmonized AI liability regime within the EU would highly be necessary.
“Member States have not yet adopted specific legislation related to the regulation of liability for AI, with some exceptions related to drones, autonomous vehicles and medical AI applications. Timely action at EU level would therefore reduce regulatory fragmentation and costs for producers of AI while also helping to secure high levels of protection for fundamental and consumer rights in the EU’’