PURPOSE: to promote the rollout of trustworthy artificial intelligence (AI) by ensuring that victims of damage caused by AI obtain equivalent protection to victims of damage caused by products in general.

PROPOSED ACT: Directive of the European Parliament and of the Council.

ROLE OF THE EUROPEAN PARLIAMENT: the European Parliament decides in accordance with the ordinary legislative procedure on an equal footing with the Council.

BACKGROUND: according to a representative survey carried out in 2020, liability ranked amongst the top three barriers to the use of AI by European companies.

Current national liability rules, in particular based on fault, are not suited to handling liability claims for damage caused by AI-enabled products and services . Under such rules, victims need to prove a wrongful action or omission by a person who caused the damage.

Given the specific characteristics of AI, it may be cumbersome or too costly for victims to identify the person responsible and to prove the conditions required to win their case. As a result, victims may be deterred from seeking compensation.

The national AI strategies show that several Member States are considering, or even concretely planning, legislation on civil liability for AI. Therefore, if the EU does not act, it is expected that Member States will adapt their national liability rules to the challenges of AI.

In the absence of EU harmonised rules for the compensation of damages caused by AI systems, providers, operators and users of AI systems on the one hand, and injured persons on the other, would be faced with 27 different liability regimes, resulting in different levels of protection and distorting competition between companies in different Member States.

In its White Paper on AI of 19 February 2020, the Commission undertook to promote the uptake of AI and addressed the risks associated with some of its uses by fostering excellence and trust. In the AI liability report accompanying the White Paper, the Commission identified specific challenges posed by AI to existing liability rules.

CONTENT: the purpose of this Directive is to improve the functioning of the internal market by laying down uniform requirements for certain aspects of non-contractual civil liability for damage caused with the involvement of AI systems . It follows the legislative initiative resolution in which the European Parliament called on the Commission to adopt a proposal for a civil liability regime for AI and adapt private law to the needs arising from the transition to the digital economy.

The proposal applies to non-contractual civil law claims for damages caused by an AI system, where such claims are brought under fault-based liability regimes. This means namely regimes that provide for a statutory responsibility to compensate for damage caused intentionally or by a negligent act or omission.

The proposed directive alleviates the burden of proof in a very targeted and proportionate manner through the use of disclosure and rebuttable presumptions.

A right of access to evidence

The proposed Directive aims to provide persons seeking compensation for damage caused by high-risk AI systems with effective means to identify potentially liable persons and relevant evidence for a claim. At the same time, such means serve to exclude falsely identified potential defendants.

Under the Directive, a court could order the disclosure of relevant evidence concerning specific high-risk AI systems suspected of causing damage. Requests for evidence should be addressed to the provider of an AI system, the person subject to the provider’s obligation or the user. Requests should be supported by facts and evidence sufficient to establish the plausibility of the contemplated claim for damages and the requested evidence should be at the addressees’ disposal.

By limiting the obligation to disclose or preserve to necessary and proportionate evidence, the proposal aims to limit disclosure to the minimum necessary and to prevent general requests. On the other hand, disclosure would be subject to appropriate safeguards to protect sensitive information, such as trade secrets.

Presumption of causal link in the case of fault

It may be challenging for the claimant to establish a causal link between the failure to comply with a duty of care under EU or national law and the result of the AI system or the failure of the AI system to produce an output that gave rise to the relevant damage.

Therefore, the proposal provides for a targeted rebuttable presumption regarding this causal link. The rebuttable presumptions will give those seeking compensation for damage caused by AI systems a more reasonable burden of proof and a chance to succeed with justified liability claims.

The fault of the defendant must be proven by the claimant according to the applicable Union or national rules. Such fault can be established, for example, for non-compliance with a duty of care pursuant to the AI Act or pursuant to other rules set at Union level, such as those regulating the use of automated monitoring and decision-making for platform work or those regulating the operation of unmanned aircraft.

Such fault can also be presumed by the court on the basis of a non-compliance with a court order for disclosure or preservation of evidence. Still, it is only appropriate to introduce a presumption of causality when it can be considered likely that the given fault has influenced the relevant AI system output or lack thereof, which can be assessed on the basis of the overall circumstances of the case. At the same time, the claimant still has to prove that the AI system (i.e. its output or failure to produce one) gave rise to the damage.