Over the past three years the EU has become involved in developing legislation to deal with the operation of Artificial Intelligence (AI) in the Union. There are three strands to this legislation: the overall regulatory AI Act; the updating of the 1986 Product Liability Directive; addressing civil liability arising out of the operation of AI systems. On 22 October 2020 the European Parliament sent a draft regulation to the Commission for a new strict liability regime for operators of AI systems. The Parliament’s proposal was followed on 28 September 2022 by the European Commission’s proposal for the AI Liability Directive along with a proposed updating of the 1986 Product Liability Directive. This blog pointed out that the proposed Regulation could lead to a confusing overlap with maritime strict liability regimes in the context of vessels at MASS 3 and 4.
Unlike the Parliament’s proposal of October 2020, the Commission’s proposal is framed as a Directive, and contains no substantive rules regarding liability arising out of use of an AI system. Instead, the proposed Directive applies to non-contractual fault-based civil law claims for damages, in cases where the damage caused by an AI system occurs after the end of the transposition period, but the Directive lays down two sets of common rules.
First, Article 3 deals with the disclosure of evidence on high-risk artificial intelligence (AI) systems to enable a claimant to substantiate a non-contractual fault-based civil law claim for damages.
Second, Article 4 deals with the burden of proof in establishing causality in non-contractual fault-based civil law claims brought before national courts for damages caused by an AI system. For the presumption of causality to apply, the fault of the defendant should be established as a human act or omission which does not meet a duty of care under Union law or national law that is directly intended to protect against the damage that occurred. It should also be necessary to establish that it can be considered reasonably likely, based on the circumstances of the case, that the fault has influenced the output produced by the AI system or the failure of the AI system to produce an output and the claimant should still be required to prove that the output or failure to produce an output gave rise to the damage.
However, fault still has to be proved under the applicable Union or national laws, although fault can be established in respect of non-compliance with Union rules which specifically regulate high-risk AI systems. It is likely that in the future such rules will apply to vessels at MASS 3 and 4 for entry into ports and the territorial sea of Member States. The Directive does not affect rules of Union law regulating conditions of liability in the field of transport. With maritime transport the only such rules of Union law concerning fault based civil law claims would be Directive 2009/20/EC on the insurance of shipowners for maritime claims.
Art 5 provides for the Commission to submit a report to the Parliament, the Council, and the Economic and Social Committee, assessing the Directive’s achievement five years after its transposition. In particular, that review should examine whether there is a need to create no-fault liability rules for claims against the operator combined with a mandatory insurance for the operation of certain AI systems, as suggested by the European Parliament resolution of 20 October 2020 on a civil liability regime for artificial intelligence.
The restriction to fault-based liability regimes means that, in relation to MASS 3 and 4 vessels operating with the territory of the Union, the proposed Directive will have no application to the two current strict liability pollution regimes, the CLC and the Bunkers Convention, and will have no application to the HNS regime when it eventually comes into force. It will, though, have application in the Member States to fault based tort claims such as general pollution claims and collision claims, as regards the rebuttable presumption of a causal link in the case of fault provided for in Art 4, and almost certainly as regards the evidential provisions in Art 3 if MASS 3 and 4 vessels are eventually classified as ‘high risk’.
The proposed Directive now has to go back to the Parliament and the Council, and may well be subject to amendment. The European Economic and Social Committee (EESC) adopted an opinion on the proposal on 25 January 2023 broadly welcoming the proposal but insisting upon clear legal definitions, calling upon the Commission to closely monitor the development of financial guarantees or insurance covering AI liability and recommending the Directive be reviewed three years after entry into force.