EPRS BRI(2023)739342 EN · Page 10

EPRS | European Parliamentary Research Service 
10 
The Ada Lovelace Institute recommends introducing a strict liability regime for high-risk AI systems 
and a complete reversal of the burden of proof for other AI systems to clarify the scope and 
application and offer greater legal certainty.31 Another commentator stresses that the draft text 
does not make it easy for injured parties to establish a presumption of causality, given the heavy 
burden of proof they face, and argues that amendments should be introduced to facilitate the 
mechanism of redress available to victims of AI harm.  
Burden of proof and effectiveness 
Some commentators highlight that it is questionable whether a fault-based liability regime would 
succeed in simplifying victims' claims even with a regime of presumption of causality. They argue 
that AI systems can be so complex that even when a user complies with their duty of care, damage 
can still arise and it is not clear who will be held liable for such damages and on what grounds.32 
Impact on innovation 
According to some experts, providers of AI systems will find it difficult to adequately protect 
themselves from liability, as they will have to comply with several product safety and liability 
regulations, including potential claims under the new AI liability directive and the PLD, and the 
forthcoming AI act. As a result, there is a risk of substantial chilling effect on AI innovation in 
Europe.33 
Interplay between EU and nation rules 
The Commission chose to table a directive. However, it has been stressed that if Member States 
maintain a strict liability regime in relation to certain cases, this would run against providing 
economic operators with legal certainty, thereby threatening the harmonisation of liability rules on 
a European scale.34 Furthermore, since the directive does not harmonise the conditions related to 
the damage, discrepancies between Member States' national systems can be a source of problems. 
For instance in some Member States, one would be able to invoke the directive as a means of 
claiming compensation due to a psychological harm caused by an AI system, while in other Member 
States, such harms would not be eligible for compensation.35  
Legislative process 
In Parliament, the file has been assigned to the Legal Affairs Committee (JURI) and Axel Voss (EPP, 
Germany) has been appointed as rapporteur. The next step for the directive is for the European 
Parliament and Council to consider and adopt the draft text.  
 
Debate on the legal personality for AI systems. In its landmark 2017 resolution on Civil Law Rules on 
Robotics, the Parliament called on the Commission to explore the possibility to grant the most sophisticated 
autonomous robots a status of electronic persons (close to the notion of 'legal personality'), so that such 
robots can be held responsible for any damage they may cause. A number of academics in the field have 
refuted the need to adopt such an approach and grant legal personality to emerging digital technologies.i 
In this respect, it has been stressed that granting AI a legal personality would result in shifting liabilities to 
AI systems and in shielding humans from the consequences of their conduct. However, for other experts, 
adapting traditional policies on strict liability and fault-based liability to the context of AI services and 
products as proposed is likely to fall short, especially when no human is liable for damages or when the 
victim i