I. Introduction
On 2 October 1997, the Member States of the European Union (EU) signed the Treaty of Amsterdam and endowed the European legislature with a competence in the field of private international law that is now found in Article 81(2)(c) of the Treaty on the Functioning of the European Union.Footnote 1 In the following two decades, the EU created an expanding body of private international law.Footnote 2 In particular, the Rome II Regulation on the law applicable to non-contractual obligations was enacted on 11 July 2007.Footnote 3 Only eleven months later, the Rome I Regulation on the law applicable to contractual obligations was adopted.Footnote 4 Although both Regulations are already rather comprehensive, gaps as well as inconsistencies remain.Footnote 5 In light of the rapid technological development since 2009, the issue as to whether there is a need for specific rules on the private international law of artificial intelligence (AI) has to be addressed.Footnote 6 After the European Parliament’s JURI Committee had presented a proposal for a civil liability regime for AI in April 2020,Footnote 7 the European Parliament adopted – with a large margin – a pertinent resolution with recommendations to the Commission on 20 October 2020.Footnote 8 This resolution is part of a larger regulatory package on issues of AI.Footnote 9 The draft regulation (DR) proposed in this resolution is noteworthy not only with regard to the rules on substantive law that it contains,Footnote 10 but also from a choice-of-law perspective because it introduces new, specific conflicts rules for AI-related aspects of civil liability.Footnote 11 In the following contribution, I analyse and evaluate the European Parliament’s proposal against the background of the already existing European regulatory framework on private international law, in particular the Rome I and II Regulations.
II. The Current European Framework
1. The Goals of PIL Harmonisation
The basic economic rationale underlying the Rome II Regulation is succinctly captured in its Recital 6, which reads as follows:
The proper functioning of the internal market creates a need, in order to improve the predictability of the outcome of litigation, certainty as to the law applicable and the free movement of judgments, for the conflict-of-law rules in the Member States to designate the same national law irrespective of the country of the court in which an action is brought.
This Recital epitomises the basic tenet of the methodology developed by Friedrich Carl von Savigny in the nineteenth century, in other words, the goal of international decisional harmony.Footnote 12 The Commission’s explanation for its Rome II draft of 2003 is even more explicit with regard to the deterrence of forum shopping: unless conflicts rules for non-contractual obligations become unified, ‘[t]he risk is that parties will opt for the courts of one Member State rather than another simply because the law applicable in the courts of this State would be more favourable to them.’Footnote 13 The explanation for the draft of 2003 also makes clear that a unification of tort conflicts rests on a sound economic rationale, the reduction of transaction costs borne by the parties. A European Regulation on tort conflicts ‘allows the parties to confine themselves to studying a single set of conflict rules, thus reducing the cost of litigation and boosting the foreseeability of solutions and certainty as to the law.’Footnote 14 This rationale is particularly important for tort conflicts, because, contrary to contract conflicts, a choice of the applicable law ex ante was traditionally not available in many jurisdictions.Footnote 15 Even if the parties enjoy that possibility, they will frequently not be able to exercise this right because they do not anticipate an accident to happen.Footnote 16 Accordingly, clear objective conflicts rules have significantly greater weight in tort than in contract cases.Footnote 17 This is an important factor facilitating the emergence of new technologies with cross-border implications, such as driverless cars.Footnote 18
Moreover, the force of a practical example that would emanate from a successful codification of European conflicts rules on AI must not be underestimated. Although the initial American reaction towards the Rome II Regulation was rather critical, denouncing the final text as a ‘missed opportunity’ to transplant US doctrines to Europe,Footnote 19 there is a palpable transatlantic interest in recent European developments and the lessons that these may hold for the United States.Footnote 20 A well-known American conflicts scholar even recommended the European codification of tort conflicts as a model for further US legislation.Footnote 21 While the ‘end of history’ for private international law (i.e. a full convergence of US and European conflict of laws in torts),Footnote 22 is still a long road ahead, a successful EU legislation on the law applicable to liability issues of AI will certainly increase the prospects for creating harmonised conflicts rules in this area on a global level.
2. The Subject of Liability
Both the Rome I and II Regulations only address the liability of natural personsFootnote 23 and ‘companies and other bodies, corporate or unincorporated’.Footnote 24 Thus, the question arises as to whether an AI system could be classified as another ‘unincorporated body’ within the meaning of these provisions.Footnote 25 There is a parallel discussion about attributing legal personality to AI-systems in substantive private law.Footnote 26 Although the mere wording of the English version of the Rome I and II Regulations would arguably allow such an innovative interpretation, other linguistic versions suggest a narrower, more traditional reading of the Regulations (e.g. the German one, which speaks of ‘Gesellschaften, Vereine und juristische Personen’). Since the law applicable to legal personality is not yet determined by EU private international law, but remains subject to domestic choice-of-law rules within the boundaries of the freedom of establishment,Footnote 27 it would be unwise to burden the Rome I and II Regulations with a regulatory aspect that is, from the point of view of international contract and tort law, merely an incidental question. Thus, the law applicable to legal personality will have to be determined by other measures, e.g. by a regulation based on the draft presented by the European Group for Private International Law in 2016.Footnote 28
3. Non-Contractual Obligations: The Rome II Regulation
a. Scope
The Rome II Regulation determines the law applicable to non-contractual obligations, in particular torts. The notion of ‘non-contractual obligation’ must be interpreted as an autonomous concept.Footnote 29 It covers both strict and fault-based liability.Footnote 30 Generally speaking, all types of harm or damage are covered, such as physical damage to property, pure economic loss, and immaterial harm.Footnote 31 The Rome II Regulation is limited to civil and commercial matters;Footnote 32 notably, it does not cover the liability of the state for acts and omissions in the exercise of state authority.Footnote 33 Thus, the law applicable to a Member State’s liability for the use of AI for the purpose of international police surveillance or military operations, for example, is determined by domestic choice-of-law rules.Footnote 34 Moreover, the Rome II Regulation is not applicable to non-contractual obligations arising out of violations of privacy and rights relating to personality, including defamation.Footnote 35 Therefore, the law applicable to any kind of use of AI that violates a person’s right to privacy or causes damage to their reputation must still be determined by domestic choice-of-law rules, such as Articles 40–42 of the German EGBGB.Footnote 36 Finally, although the rules of the Rome II Regulation are of European origin, they shall be applied whether or not the law specified by them is the law of an EU Member State.Footnote 37 Thus, according to this principle of ‘universal application’, even if an AI system operated by a British company causes damage to a person in Switzerland, the court of an EU Member State will determine the law applicable to such a case pursuant to the Rome II Regulation.Footnote 38
b. The General Rule (Article 4 Rome II)
The basic rule for torts in general is found in Article 4(1) Rome II, which refers to the place of injury. Recital 15 Rome II acknowledges that ‘lex loci delicti is the basic solution for non-contractual obligations in virtually all the Member States’. Nevertheless, the diverging interpretations of this principle by various Member States’ legislatures and courts in complex cases (place of injury, place of acting, or even both under the so-called theory of ubiquity) had in the past led to considerable legal uncertainty.Footnote 39 The preference for the place of injury is justified because, generally speaking, it strikes ‘a fair balance’ between the interest of the person claimed to be liable to foresee the applicable law and the interests of the person sustaining the damage.Footnote 40 From an economic point of view, the place of injury will usually lead to a fair distribution of the costs for obtaining the relevant legal information: In most cases, the person claimed to be liable should be able to anticipate that his or her acts may cause harm in another country, whereas the victim should be able to rely on the legal standard of the environment to which he or she exposed his or her body or property.Footnote 41 While the tortfeasor is thus forced to internalise the costs for negative externalities arising in other countries,Footnote 42 the victim is given the opportunity to structure his or her insurance in accordance with the law to which he or she is presumably accustomed.Footnote 43 Since Article 4(1) Rome II is based on the idea of striking ‘a fair balance’ between the alleged tortfeasor and victim, this neutral provision must not be interpreted in a one-sided fashion that favours the plaintiff. The Rome II Regulation does not, as a general principle, embrace the plaintiff-friendly principle of ubiquity found in German or Italian private international law.Footnote 44
The Rome II Regulation contains a significant number of specific rules for special torts.Footnote 45 This considerably reduces the weight that the general rule has to carry, which applies only ‘unless otherwise provided for in this Regulation’.Footnote 46 The main group of cases of practical importance that are exclusively governed by the general rule instead of specific rules are traffic accidents.Footnote 47 However, even in this regard, the scope of application of Article 4 Rome II is limited in practice. The full communitarisation of private international law is impeded by the fact that there already exist two supranational instruments dealing with important areas of tort conflicts, namely, the Hague Convention on the law applicable to Traffic Accidents (HCTA) and the Hague Convention on the law applicable to Products Liability (HCP).Footnote 48 Both conventions count several EU Member States among their parties.Footnote 49 Those Member States were (and are) unwilling to withdraw from the respective conventions.Footnote 50 Since the EU could arguably not terminate their membership without their consent, rules governing the collision between EU conflicts rules and the Hague conventions had to be invented.Footnote 51 The solution finally codified in the Rome II Regulation provides that the Regulation does not prejudice the application of existing conventions that contain conflicts rules for non-contractual obligations.Footnote 52 The Rome II Regulation takes precedence, however, over conventions concluded exclusively between two or more of them insofar as such conventions concern matters governed by the Regulation.Footnote 53 Since both pertinent Hague conventions have a sizeable number of non-EU state parties, this exception is of little practical use.Footnote 54 Even if a traffic accident is only connected with, for example, France and Germany, French courts have to apply the HCTA, whereas a German court must determine the applicable law under the Rome II Regulation.Footnote 55 Thus, in two of the most important areas of tort conflicts, traffic accidents and product liability, European private international law remains fragmented and continues to offer ample possibilities of forum shopping.Footnote 56 This situation is exacerbated by the fact that the Rome II Regulation excludes the possibility of renvoi.Footnote 57 Thus, cases involving driverless cars, for example, may be subject to different laws in various Member States.Footnote 58
The lex loci damniFootnote 59 is displaced in cases where the person claimed to be liable and the person sustaining the damage both have their habitual residence in the same country at the time when the damage occurs.Footnote 60 This rule had been familiar to many European codifications already before Rome II was enacted.Footnote 61 Again, it is a legitimate expression of the basic economic rationale underlying the Regulation: ‘[I]n most cases the common residence rule guarantees lower litigation costs, more efficient court administration, and international harmony of decisions’.Footnote 62 Usually, parties who share a common habitual residence will litigate in the country where they live; moreover, their insurance coverage will, in most cases, be structured according to the standards prevailing in this country.Footnote 63
Article 4(1) and (2) Rome II are coupled with an escape clause that is meant to provide for a sufficient degree of judicial discretion in the individual case.Footnote 64 The final paragraph, which is rather an open-ended standard than a rule, combines a fairly general approach in its first sentence (manifestly closer connection) with a particular example of such a connection (relationship between the parties, for example, a contract) in its second sentence. As Recital 14 Rome II shows, the drafters of the Regulation were mindful of the tension between ‘the requirement of legal certainty’ on the one hand and the ‘need to do justice in individual cases’ on the other. The Recital explains that
this Regulation provides for a general rule but also for specific rules and, in certain provisions, for an ‘escape clause’ which allows a departure from these rules where it is clear from all the circumstances of the case that the tort/delict is manifestly more closely connected with another country. This set of rules thus creates a flexible framework of conflict-of-law rules. Equally, it enables the court seised to treat individual cases in an appropriate manner.
Finally, Article 14 Rome II provides for a modern and liberal approach to party autonomy for non-contractual obligations, allowing a choice of the applicable law both ex post and, provided certain conditions are met, ex ante.Footnote 65 The reasons for this liberal approach are spelled out in the first sentence of Recital 31: ‘To respect the principle of party autonomy and to enhance legal certainty, the parties should be allowed to make a choice as to the law applicable to a non-contractual obligation.’ Party autonomy enhances legal certainty in two ways.Footnote 66 First, the flexible approach of the Regulation, which is characterised by a rather generous array of escape clauses,Footnote 67 introduces a potential source of litigation that must be balanced by giving parties the possibility of quickly resolving any dispute on the applicable law.Footnote 68 Secondly, the substantive laws of the Member States are characterised by significant divergences as far as the proper boundaries between tort and contract law are concerned. This is particularly true for cases such as pre-contractual liability, liability for pure economic loss, and the protection of third persons who are not a party to an existing contract with the person claimed to be liable.Footnote 69 Thus, parties who want to avoid a protracted litigation on issues of classification are well advised to choose the law applicable not only to their contractual obligations, but also to their non-contractual obligations.Footnote 70
c. The Rule on Product Liability (Article 5 Rome II)
With regard to product liability, Article 5 Rome II strives to create a balance between an effective protection of the victim, who is often a consumer and typically regarded as the weaker party, on the one hand, and the producer’s interest in foreseeability of the applicable law, on the other.Footnote 71
Article 5(1) Rome II presupposes a damage ‘caused by a product’. The notion of ‘product’ must be interpreted autonomously;Footnote 72 the Commission’s Explanatory Memorandum of 2003Footnote 73 refers to the definition found in the EU Directive on Product Liability.Footnote 74 The substantive EU law on product liability so far only applies to physical goods.Footnote 75 Thus, strict liability for data processing cannot be based on the current Product Liability Directive.Footnote 76 A working group hosted by the European Law Institute has recently published a paper on giving the Product Liability Directive a digital ‘update’, but this reform process is still in its first stages.Footnote 77 Although the rules of the current Product Liability Directive may be extended to cover standard software delivered on a DVD, for example,Footnote 78 it is controversial whether software that was designed to meet the specific needs of the customer could be classified as a ‘product’.Footnote 79 Those delineations are generally transferred to Article 5(1) Rome II.Footnote 80 In cases of autonomous driving, however, the software will be sold as an integral part of a car. In cases where software is embedded in a physical good, both the Product Liability Directive and Article 5(1) Rome II apply.Footnote 81
The cascade of connections found in Article 5 Rome II is structured as follows: first, parties may choose the law applicable to product liability claims under the general provision on party autonomy.Footnote 82 Likewise, the Rome II Regulation provides for an accessory connection of product liability claims to a pre-existing relationship, such as a contract, between the parties.Footnote 83 Both steps constitute major improvements compared to the Hague Convention on the law applicable to product liability,Footnote 84 which failed to include such rules.
Secondly, if both parties have their habitual residence in the same country, the law of that state applies.Footnote 85
Thirdly, if none of the above applies, Article 5(1) Rome II basically refers to the law of the state where the product was marketed, provided that the place of marketing coincides with one of three other territorial factors (the victim’s habitual residence, the place where the product was acquired, the place of injury) and that the person claimed to be liable (usually the producer) could reasonably foresee the marketing of the product or a product of the same type in this country. Contrary to specific provisions on product liability, for example in ItalyFootnote 86 or Switzerland,Footnote 87 Article 5(1) Rome II is not an alternative connection, but ranks the connecting factors in a hierarchical order. Firstly, the law applicable is that of the victim’s habitual residence, provided that (1) it coincides with the place of marketing and (2) the producer does not succeed at proving that he could not foresee the marketing of this or a similar product in this country.Footnote 88 If one of those conditions (marketing, foreseeability) is not met, the law of the country in which the product was acquired applies, again subject to a coincidence with the place of marketing and the test of foreseeability.Footnote 89 If the applicable law cannot be determined at this stage, the law of the country in which the ‘damage [read: injury] occurred’, applies, if at least in this country the two additional requirements (marketing, foreseeability) are met.Footnote 90 If all of the three countries enumerated in Article 5(1) Rome II do not pass the test of foreseeability, the applicable law is that of the producer’s habitual residence.
This rather unwieldy ‘cascade system of connecting factors’Footnote 91 fails to achieve wholly convincing results. First, even after the Rome II Regulation has been in force now for more than a decade, it has not induced a single Member State, which is a party to the HCP, to denounce this convention. On the contrary, under Article 28 Rome II, the HCP takes precedence over the Rome II Regulation. The result is that, since 2009, Europeans have two different regimes on product liability conflicts which are both influenced by a similar methodology (grouping of contacts), but which do not yield uniform results in practice.
While Recital 20 explains that the ‘conflict-of-law rule in matters of product liability should meet the objectives of fairly spreading the risks inherent in a modern high-technology society, protecting consumers’ health, stimulating innovation, securing undistorted competition and facilitating trade,’ it must be kept in mind that Article 5(1) Rome II is not limited to business-to-consumer (B2C) cases, but applies to business-to-business (B2B) cases as well.
Since the connecting factor that enjoys primacy in the basic ruleFootnote 92 is relegated to the last rung of the ladder in cases of product liability,Footnote 93 drawing the line between general tortious liability and product liability is decisive in traffic accidents involving autonomous cars.Footnote 94 Thus, one may argue that there is a need for a special conflicts rule for those cases. A further complication arises from the above-mentioned fact that, in quite a number of member states, the law applicable to traffic accidents or product liability is still not determined by the Rome II Regulation, but by the pertinent Hague Conventions of the early 1970s (see Sub-section II.3(b)). Therefore, even an amendment to the Rome II Regulation would not create European legal unity in this regard.
d. Special Rules in EU Law (Article 27 Rome II)
Pursuant to Article 27 Rome II, special EU conflicts rules take precedence over Rome II. In particular, the conflicts rules of the General Data Protection RegulationFootnote 95 may be relevant in cases involving AI.Footnote 96 In the course of the preparation of the Rome II Regulation, industry lobbies argued for codifying the ‘country of origin’-approach as a choice-of-law rule.Footnote 97 While those attempts failed, Article 27 Rome II explicitly states that ‘provisions of Community law which, in relation to particular matters, lay down conflict-of-law rules relating to non-contractual obligations’ take precedence over the Regulation. Moreover, Recital 35 Rome II adds that the Regulation:
should not prejudice the application of other instruments laying down provisions designed to contribute to the proper functioning of the internal market insofar as they cannot be applied in conjunction with the law designated by the rules of this Regulation. The application of provisions of the applicable law designated by the rules of this Regulation should not restrict the free movement of goods and services as regulated by Community instruments, such as … [the] Directive on electronic commerce[Footnote 98].
The precise reach of this exhortation is hard to define because the Directive on electronic commerce itself takes the somewhat schizophrenic position that it does not contain conflict-of-law rules,Footnote 99 while at the same time laying down the country-of-origin principle in its Article 3(1) and (2).Footnote 100 With regard to violations of rights of personality, a field not covered by Rome II, the CJEU tried to clarify matters as follows:Footnote 101
Article 3 of Directive 2000/31/EC of the European Parliament and of the Council of 8 June 2000 on certain legal aspects of information society services, in particular electronic commerce, in the Internal Market (“Directive on electronic commerce”), must be interpreted as not requiring transposition in the form of a specific conflict-of-laws rule. Nevertheless, in relation to the coordinated field, Member States must ensure that, subject to the derogations authorized in accordance with the conditions set out in Article 3(4) of Directive 2000/31, the provider of an electronic commerce service is not made subject to stricter requirements than those provided for by the substantive law applicable in the Member State in which that service provider is established.
If the European legislature were to codify special conflicts rules on AI, such a regulation would not only supersede the Rome II Regulation pursuant to its Article 27, but arguably also take precedence over the Hague Conventions. The respective Articles 15 of the HCTA and the HCP state that the Hague Conventions shall not prevail over other Conventions ‘in special fields’ to which the contracting states are or may become parties. Although an EU Regulation is surely not a ‘convention’ within the technical meaning of those provisions, one may argue that Article 15 HCTA/HCP should apply by way of an analogy to any EU Regulation dealing with the law applicable to autonomous driving, for example.
4. Contractual Obligations: The Rome I Regulation
a. Scope
Complementing Rome II, the Rome I Regulation determines the law applicable to contractual obligations.Footnote 102 Mirroring the Rome II Regulation,Footnote 103 the notion of contractual obligation must be interpreted as an autonomous concept.Footnote 104 Thus, the Rome I Regulation designates the law applicable to so-called smart contracts, for example.Footnote 105 Likewise, the Rome I Regulation is of universal application as well.Footnote 106
b. Choice of Law (Article 3 Rome I)
Party autonomy is largely permitted by Article 3 Rome I.Footnote 107 Consumers, however, must not be deprived of the protection accorded to them by the law of their habitual residence.Footnote 108
c. Objective Rules (Articles 4 to 8 Rome I)
Usually, the habitual residence of the service provider determines the law applicable to contracts for services.Footnote 109 With regard to consumers, the law of the consumer’s habitual residence applies under the conditions set out in Article 6(1) Rome I.Footnote 110
d. Special Rules in EU Law (Article 23 Rome I)
Special conflicts rules in other EU legal instruments prevail over the Rome I Regulation.Footnote 111 There are occasional conflicts rules in older consumer directives;Footnote 112 however, the more recent directive on digital content and services does not contain any such rule.Footnote 113 On the contrary, Recital 80 of said directive explicitly states that ‘[n]othing in this Directive should prejudice the application of the rules of private international law, in particular Regulations (EC) No 593/2008 and (EU) No 1215/2012 of the European Parliament and of the Council’.
III. The Draft Regulation of the European Parliament
1. Territorial Scope
With regard to substantive law, the draft regulation distinguishes between legally defined high-risk AI-systemsFootnote 114 and other AI-systems involving a lower riskFootnote 115. For high-risk AI-systems, the draft regulation would introduce an independent set of substantive rules providing for strict liability of the system’s operator.Footnote 116 Further provisions deal with the amount of compensation,Footnote 117 the extent of compensationFootnote 118 and the limitation period.Footnote 119 The spatial scope of those autonomous rules on strict liability for high-risk AI-systems is determined by Article 2 DR, which reads as follows:
1. This Regulation applies on the territory of the Union where a physical or virtual activity, device or process driven by an AI-system has caused harm or damage to the life, health, physical integrity of a natural person, to the property of a natural or legal person or has caused significant immaterial harm resulting in a verifiable economic loss.
2. Any agreement between an operator of an AI-system and a natural or legal person who suffers harm or damage because of the AI-system, which circumvents or limits the rights and obligations set out in this Regulation, concluded before or after the harm or damage occurred, shall be deemed null and void as regards the rights and obligations laid down in this Regulation.
3. This Regulation is without prejudice to any additional liability claims resulting from contractual relationships, as well as from regulations on product liability, consumer protection, anti-discrimination, labour and environmental protection between the operator and the natural or legal person who suffered harm or damage because of the AI-system and that may be brought against the operator under Union or national law.
The unilateral conflicts rule found in Article 2(1) DR would prevail over the Rome II Regulation on the law applicable to non-contractual relations pursuant to Article 27 Rome II.Footnote 120 However, the Rome II Regulation still applies to additional liability claims mentioned in Article 2(3) DR. Moreover, Article 2(1) DR seems to limit the applicability of the draft regulation to cases where the harm was suffered on the territory of the European Union.Footnote 121 This stands in stark contrast with the principle of universal application that is one of the cornerstones of the Rome II Regulation.Footnote 122 If a high risk AI-system operated in Freiburg, Germany, for example, caused damage in Basel, Switzerland, the preconditions set out in Article 2(1) DR would not be met; thus, one would have to resort to the Rome II Regulation to determine the law applicable to the Swiss victim’s claims.
2. The Law Applicable to High Risk Systems
Furthermore, it must be noted that Article 2(1) DR deviates considerably from the choice-of-law framework of Rome II. While Article 2(1) DR reflects the lex loci damni approach enshrined as the general conflicts rule in the Rome II Regulation,Footnote 123 one must not overlook the fact that product liability is subject to a special conflicts rule, namely Article 5 Rome II, which is considerably friendlier to the victim of a tort than the general conflicts rule.Footnote 124 This cascade of connections is evidently influenced by the desire to protect the mobile consumer from being confronted with a law that may be purely accidental from his point of view. The lex loci damniFootnote 125 may have neither a relationship with the legal environment that consumers are accustomed toFootnote 126 nor with the place where they decided to expose themselves to the danger possibly emanating from the product.Footnote 127 The rule reflects the presumption that a defective product will affect most consumers in the country where they are habitually resident. Insofar, Article 2(1) DR is, in comparison with the Rome II Regulation, friendlier to the operator of a high-risk AI-system than to the consumer.
Even if one limits the comparison between Article 2(1) DR and the Rome II Regulation to the latter’s general rule,Footnote 128 it is striking that the DR does not adopt familiar approaches that allow for deviating from a strict adherence to lex loci damni. Contrary to Article 4(2) Rome II, where the person claimed to be liable and the person sustaining damage both have their habitual residence in the same country at the time when the damage occurs, Article 2 DR does not allow to apply the law of that country. Moreover, an escape clause such as Article 4(3) or Article 5(2) Rome II is missing in Article 2 DR. Finally, yet importantly, Article 2(2) DR bars any party autonomy with regard to strict liability for a high-risk AI-system, which deviates strongly from the liberal approach found in Article 14 Rome II.
3. The Law Applicable to Other Systems
Apart from the operator’s strict liability for high-risk AI-systems, the draft regulation would introduce a fault-based liability rule for other AI-systems.Footnote 129 In principle, the spatial scope of the latter liability rule would also be determined by Article 2 DR as already described.Footnote 130 However, unlike the comprehensive set of rules on strict liability for high-risk systems, the draft regulation’s model of fault-based liability is not completely autonomous. Rather, the latter type of liability contains important carve-outs regarding the amounts and the extent of compensation as well as the statute of limitations. Pursuant to Article 9 DR, those issues are left to the domestic laws of the Member States. More precisely, Article 9 DR states: ‘Civil liability claims brought in accordance with Article 8(1) shall be subject, in relation to limitation periods as well as the amounts and the extent of compensation, to the laws of the Member State in which the harm or damage occurred.’ Thus, we find a lex loci damni approach with regard to fault-based liability as well. Again, the principle of universal applicationFootnote 131 is discarded; contrary to the rules of Rome II, Article 9 DR is a unilateral conflicts rule that only refers to ‘the laws of the Member State in which the harm or damage occurred’. Moreover, all the modern approaches codified in the Rome II Regulation – the cascade of connecting factors for product liability claims, the common habitual residence rule, the escape clause, and party autonomy – are strikingly absent from Article 9 DR as well.
Finally, yet importantly, Article 9 DR leads to a split between the law applicable to the basis of liability, on the one hand, and the law applicable to limitation periods as well as the extent of compensation, on the other. This dépeçage stands in stark contrast with the general scope that Article 15 Rome II assigns to the lex causae. Pursuant to Article 15(a) Rome II, the law applicable to a non-contractual obligation under the Rome II Regulation covers both the basis and the extent of liability.Footnote 132 In addition, Article 15(h) Rome II provides that the law designated by the Rome II Regulation also applies to rules of prescription and limitation.Footnote 133 As Axel Halfmeier explains, ‘the general tendency of the [Rome II] Regulation is to expand the reach of the lex causae and limit the role of the lex fori [because] the goal of the Rome Regulations is to produce harmony in results among the Member States’ courts’Footnote 134 – the classic Savignyan goal of international decisional harmony mentioned above.Footnote 135 Of course, one has to take into account that Article 9 DR does not refer to the lex fori, but to the lex loci damni. Insofar, the rule does not offer any incentive for forum shopping. However, the unitary approach underlying Article 15 Rome II also serves the goal of ‘avoiding the risk that the tort or delict is broken up in to several elements, each subject to a different law’.Footnote 136 Insofar, Article 15 Rome II aims at preventing the ‘legal uncertainty’ associated with applying different laws to a single case.Footnote 137 Particularly with regard to Article 15(h) Rome II, the Court of Justice of the EU (CJEU) ‘pointed out that, in spite of the variety of national rules of prescription and limitation, Article 15(h) of the Rome II Regulation expressly makes such rules subject to the general rule on determining the law applicable’.Footnote 138 Creating a dépeçage between an autonomous rule on the conditions of liability, on the one hand, and the law applicable to the extent of damages and prescription issues, on the other, may lead to difficult questions of characterisation and adaptation. For example, the question may arise which particular rule of prescription of the lex loci damni shall apply if the latter law comprises various types of fault-based liability or calibrates the length of the prescription period depending on the degree of fault. In such a scenario, the court addressed would have to determine which domestic type of liability most closely corresponds to the model found in Article 8 DR – a task that may not be easy to fulfil. With regard to legal policy, it is hardly convincing to subject the issue of prescription to domestic laws because the periods codified in the Member States’ laws have been criticised as being too short in light of the complexities of international cases.Footnote 139
4. Personal Scope
The draft regulation, in principle, limits its personal scope to the liability of the operator alone.Footnote 140 Recital 9 of the resolution explains that the European Parliament
[c]onsiders that the existing fault-based tort law of the Member States offers in most cases a sufficient level of protection for persons that suffer harm caused by an interfering third party like a hacker or for persons whose property is damaged by such a third party, as the interference regularly constitutes a fault-based action; notes that only for specific cases, including those where the third party is untraceable or impecunious, does the addition of liability rules to complement existing national tort law seem necessary.
Thus, for third parties, the conflicts rules of Rome II would continue to apply.
IV. Evaluation
At first impression, it seems rather strange that a regulation on a very modern technology – AI – should deploy a conflicts approach that seems to have more in common with Joseph Beale’s First Restatement of the 1930sFootnote 141 than with the modern and differentiated set of conflicts rules codified by the EU itself at the beginning of the twenty-first century (i.e. the Rome II Regulation). While the European Parliament’s resolution, in its usual introductory part, diligently enumerates all EU regulations and directives dealing with substantive issues of liability, the Rome II Regulation is not mentioned once in the Recitals. One wonders whether the members of Parliament were aware of the European Union’s acquis in the field of private international law at all.
V. Summary and Outlook
In April 2020, the JURI Committee of the European Parliament presented a draft report with recommendations to the Commission on a civil liability regime for AI (see Sub-section I). The draft regulation proposed therein is noteworthy from a private international law perspective because it introduces new conflicts rules for AI. In this regard, the proposed regulation distinguishes between a rule delineating the spatial scope of its autonomous rules on strict liability for high-risk AI systems (Article 2 DR), on the one hand (see Sub-section III.2), and a rule on the law applicable to fault-based liability for low-risk systems (Article 9 DR), on the other hand (see Sub-section III.3.). The latter rule refers to the domestic laws of the Member State in which the harm or damage occurred. In this chapter, I have analysed and evaluated this proposal against the background of the already existing European regulatory framework on private international law, in particular the Rome II Regulation. In sum, compared with Rome II, the conflicts approach of the draft regulation would be a regrettable step backwards in many ways. On 21 April 2021, the European Commission presented its proposal for an ‘Artificial Intelligence Act’.Footnote 142 However, this proposal contains neither rules on civil liability nor provisions on the pertinent choice-of-law issues. Thus, it remains to be seen how the relationship between the European Parliament’s draft regulation and Rome II will be designed and fine-tuned in the further course of legislation.