EPRS BRI(2023)739342 EN · Table (p.3)
From EPRS_BRI(2023)739342_EN.pdf · page 3
| 0 | 1 |
|---|---|
| Artificial intelligence liability directive | |
| claim (independent of fault), or on a claim against the producer of a defective product (victims | |
| must prove that the product was defective and the causal link between that defect and the damage). | |
| Concerns regarding existing liability rules | |
| The 2018 evaluation report of the PLD identified several shortcomings in relation to digital | |
| technologies in general and to AI in particular. | |
| First, the European Commission found that while digital content, software and data play a crucial | |
| role in the functioning of many new products, it remains unclear to what extent such intangible | |
| elements can be classified as products under the PLD. This triggers legal uncertainty regarding how | |
| injured parties can be compensated for damage caused by software, including software updates, | |
| and who will be liable for such damage. Second, the Commission's investigations showed that new | |
| technologies introduce new risks, such as openness to data | inputs that affect safety or |
| cybersecurity risks, while the PLD provides for compensation only for physical or material damage. | |
| Third, the specific characteristics of AI (e.g. opacity/lack of | transparency, explainability, |
| autonomous behaviour, continuous adaptation, limited predictability) make it particularly difficult | |
| to meet the burden of proof for a successful claim. Under current liability rules, in order to claim | |
| compensation for damages, victims usually have to prove the existence of damage, a fault of the | |
| liable person, and a causality between that fault and the damage or prove the damage, the defect | |
| and the causal link between the two. However, AI systems have characteristics that make it | |
| excessively difficult or even impossible for victims to identify and prove the fault of a potentially | |
| liable person or a defect and the causal link between that fault/defect and the damage suffered, and | |
| therefore to obtain compensation. 7 | |
| As a result, there is a risk that national courts will diverge in their approaches, which could lead to | |
| further fragmentation of liability rules for damage caused by AI across the EU. The Commission's | |
| impact assessment warns of a risk of legal fragmentation, with judges having to interpret general | |
| rules that were not designed with AI in mind to decide liability claims.8 Given that the law applicable | |
| to a liability claim in a cross-border context is by default the law of the country in which the damage | |
| occurs, different liability regimes and burden of proof rules could be applied to the same kind of AI | |
| product or service deployed in several Member States although they cause the same kind of | |
| damage. Therefore, businesses face legal uncertainty due to outdated and unclear EU and national | |
| liability rules, and victims of harm caused by AI products experience difficulty in obtaining | |
| compensation in the EU. | |
| The Commission concludes that this compensation gap could undermine citizens' trust in AI, as | |
| well as the ability of the legal and judicial system to ensure fair and equitable results in claims | |
| involving AI systems.9 | |
| Reforming the EU liability framework applicable to AI | |
| The ongoing reform of the EU liability framework applicable to AI is twofold. | |
| The Commission proposed to undertake a review of the PLD10 to adapt it to the digital age and | |
| preserve its technology-neutral nature and coverage. The proposal tabled in September 2022 aims | |
| to modernise the existing rules on the strict liability of manufacturers for defective products | |
| (from smart technology to pharmaceuticals) and ensure that victims can get fair compensation | |
| when defective products, including digital and refurbished products, cause harm. It also helps | |
| victims of damage caused by AI-enabled products to make a more effective compensation claim | |
| against the producer. As such, the PLD review concerns the adaptation of the producers' strict | |
| liability regime for defective products to allow for compensation for damages without the need | |
| to prove a fault. 11 | |
| In parallel, the European Commission has unveiled a proposal for an AI liability directive meant to | |
| tackle consumers' liability claims for damage caused by AI-enabled products and services. The |
Source Metadata