New emerging and diverse digital technologies, commonly known as “Artificial Intelligence” (AI), have brought significant changes to the environment and to our daily lives. In particular, the Internet has transformed our access to data, and other people, changing the way we perceive reality. The increasing digitalization of our lives brings with it threats and risk, causing harm and damage.
On 21.04.2021, the European Commission published a proposal for a Regulation on Liability for the Operation of Artificial Intelligence Systems (AI Regulation) setting out rules under which operators of whose digital technologies, having certain direct or indirect control over artificial intelligence system, can be held liable. All high-risk Al systems will be listed in an appendix to the AI-regulation. The concept of liability of operators under the proposed EU regulation will be strict liability for AI systems that pose a major risk to life or property, and this is similar to strict liability of manufacturers for defective (dangerous) products where they act or fail to act regardless of their intent or knowledge.
AI system operators of will not be able to evade liability by arguing that they acted with due diligence or that the harm or damage was caused by an autonomous activity, device or process driven by their AI system. However, operators can claim that they are not liable by showing that the harm or damage resulted from a force majeure event. In such a case, an operator facing a claim for compensation is required to prove that the event in question was beyond their reasonable control and unforeseeable, or, if foreseeable, could not have been prevented by taking reasonable steps. Due to the strict liability concept, force majeure can be expected to be closely scrutinized and will be given a narrow interpretation by the courts.
The AI Regulation applies in the EU, where a physical or virtual activity, device or process driven by an AI system has caused harm or damage to the life, health, or physical integrity of a natural person or to the property of a natural or legal person, or has caused significant non-pecuniary damage resulting in a verifiable economic loss. Jurisdiction might be a problem under the AI Regulation. An operator that has its place of business in an EU member state will as a rule, under civil law, fall under EU court jurisdiction. Sometimes, the location of the operator that may be liable is unclear. The AI system may have manufacturers, designers and distributors who may be the liable party, and those manufacturing companies may have no connection to the place of harm and/or damage. As liability for an AI system is also seen as tort liability, an action can be filed with the court proper for the place where the damage occurred or may occur. To sue an operator outside the EU, the operator will rather have to have minimum contacts with the state in question (purposefully avail itself of the privilege of conducting business there).
An AI system operator that has been held liable for harm or damage under the AI Regulation will be required to pay compensation: (a) up to a maximum amount of two million Euro in the event of the death of, or harm caused to the health or physical integrity of, the aggrieved party, resulting from operation of a high-risk AI system; (b) up to a maximum amount of one million Euro in the event of significant non-pecuniary damage resulting in a verifiable economic loss or of damage caused to property, including when several items of property of the aggrieved party are damaged as a result of a single operation of an AI system. Finally, civil liability claims brought under the AI Regulation, concerning harm to life, health or physical integrity, shall expire under the statute of limitations after a special period of thirty years from the date on which the damage occurred.
In view of the rapid development of digital technology, the new proposal for the European AI Regulation is aimed at unification of rules and provisions under which an AI system will operate in the future, eliminating the gaps in the current product liability regulations. It remains to be seen how the new AI law will be handled in practice by the courts and whether it will be sufficient to protect interests of societies and lives of citizens.