The future regulation on non-contractual civil liability for AI systems

By Susana Navas Navarro (Professor at the Universidad Autónoma de Barcelona)

I was surprised and struck by the fact that, after all the work carried out within the European Union (“EU”), on the subject of civil liability for Artificial Intelligence (“AI”) Systems, the European Commission has opted for a Directive (the Proposal for a Directive on adapting non contractual civil liability rules to artificial intelligence or “Proposal for a Directive”) as the instrument to regulate this issue. Moreover, a Directive with a content focused exclusively on two issues: a) the disclosure of relevant information for evidence purposes or to decide whether or not to bring forth a lawsuit and against whom (Article 3) and b) the presumption of the causal link between the defendant’s fault and the result or absence thereof that an AI system should produce (Article 4). The argument for this is the disparity of civil liability regimes in Europe and the difficulties there have always existed in harmonization (see the Explanatory Memorandum accompanying the Proposal, p. 7). Choosing a Regulation as proposed by the European Parliament[1] or the proposals of the White Paper on AI would have allowed such harmonisation, and could have included rules on evidence. It seems to me that behind this decision lies the invisible struggle, previously evidenced in other issues, between the Commission and the European Parliament. I believe that the risks for all involved in the use and handling of AI systems, especially high-risk ones, are compelling reasons in favour of harmonization and strict liability.

In relation to this aspect, the Proposal for a Directive abandons the risk-based approach that had been prevailing in this area, since it assumes that the civil liability regimes in most of the Member States are based on fault. This is referred to, for example, in Article 3(5) when presuming the breach of duty of care by the defendant or directly in Article 2(5) when defining the action for damages, or Article 4(1) when admitting the presumption of the causal link between it and the result produced by the IA system or by the absence or failure in the production of such a result which causes the damage. Therefore, if in the national civil liability regime, the case was subsumed under a strict liability regime (e.g., equated to the use or operation of a machine or vicarious liability of the employer), these rules would not apply. National procedural systems, in relation to access to evidence, are not so far r from the provisions of this future Directive.

Continue reading “The future regulation on non-contractual civil liability for AI systems”

Regulating liability for AI within the EU: Short introductory considerations


 by Francisco Andrade, Director of the Master's in Law and Informatics at UMinho
 and Tiago Cabral, Master's student in EU Law at UMinho

1. The development of Artificial Intelligence (hereinafter, “AI”) brings with it a whole new set of legal questions and challenges. AI will be able to act in an autonomous manner, and electronic “agents” will be, evidently, capable of creating changes in the legal position of natural and legal persons and even of infringing their rights. One notable example of this phenomena will be in data protection and privacy, where a data processing operation by a software agent may be biased against a specific data subject (eventually due to a faulty dataset, but also due to changes in the knowledge database of the “agent” under the influence of users or of other software agents ) and, thus, infringe the principles of lawfulness and fairness in data protection, but due to difficulties in auditing the decision one may never find why (or even find that there was a bias). More extreme examples can be arranged if we put software agents or robots in charge of matters such as making (or even assisting) decisions in court or questions related to the military.

2. One does not have to seek such extreme examples, in fact, even in entering into an agreement, a software agent may, by infringing the law, negatively affect the legal position of a person.
Continue reading “Regulating liability for AI within the EU: Short introductory considerations”

Robots and civil liability (ongoing work within the EU)


 by Susana Navas Navarro, Professor of Civil Law, Autonomous University of Barcelona

The broad interest shown by the European Union (EU) for the regulation of different aspects of robotics and artificial intelligence is nowadays very well known.[i] One of those aspects concerns the lines of thinking that I am interested in: civil liability for the use and handling of robots. Thus, in the first instance, it should be determined what is understood by “robot” for the communitarian institutions. In order to be considered as “robot”, an entity should meet the following conditions: i) acquisition of autonomy via sensors or exchanging data with the environment (interconnectivity), as well as the processing and analysis of this data; ii) capacity to learn from experience and also through interaction with other robots; iii) a minimal physical medium to distinguish them from a “virtual” robot; iv) adaptation of its behaviour and actions to the environment; v) absence of biological life. This leads to three basic categories of “smart robots”: 1) cyber-physical systems; 2) autonomous systems; 3) smart autonomous robots.[ii] Therefore, strictly speaking, a “robot” is an entity which is corporeal and, as an essential part of it, may or may not incorporate a system of artificial intelligence (embodied AI).

The concept of “robot” falls within the definition of AI, which is specified, on the basis of what scholars of computer science have advised, as: “Artificial intelligence (AI) systems are software (and possibly also hardware) systems designed by humans that, given a complex goal, act in the physical or digital dimension by perceiving their environment through data acquisition, interpreting the collected structured or unstructured data, reasoning on the knowledge, or processing the information, derived from this data and deciding the best action(s) to take to achieve the given goal. AI systems can either use symbolic rules or learn a numeric model, and they can also adapt their behaviour by analysing how the environment is affected by their previous actions. 
As a scientific discipline, AI includes several approaches and techniques, such as machine learning (of which deep learning and reinforcement learning are specific examples), machine reasoning (which includes planning, scheduling, knowledge representation and reasoning, search, and optimization), and robotics (which includes control, perception, sensors and actuators, as well as the integration of all other techniques into cyber-physical systems”.[iii]

Concerning the robot as a corporeal entity, issues related to civil liability are raised from a twofold perspective: firstly, in relation to the owner of a robot in the case of causation of damages to third parties when there is no legal relationship between them; and, secondly, regarding the damages that the robot may be caused to third parties due to its defects. From a legal standpoint, it should be noted that in most cases the “robot” is considered as “movable good” that, furthermore, may be classified as a “product”. We shall focus on each of these perspectives separately.
Continue reading “Robots and civil liability (ongoing work within the EU)”

Editorial of October 2016


by Sophie Perez Fernandes, Junior Editor

Engaging EU liability within the European Stability Mechanism framework

Last September 20th, the European Court of Justice (ECJ) delivered two judgments regarding the role of the European Commission and, to a lesser extent, the European Central Bank, in the negotiation and signing of the Memorandum of Understanding concluded between the Republic of Cyprus and the European Stability Mechanism (ESM) during the 2012-2013 financial crisis, and, in particular, in the restructuring of the banking sector in Cyprus imposed as a condition for the grant of financial assistance.

In Mallis and Malli (Joined Cases C-105/15 P to C-109/15 P), actions were brought against the European Commission and the European Central Bank for the annulment of the Eurogroup’s statement of 25 March 2013 concerning, inter alia, the restructuring of the banking sector in Cyprus. In turn, in Ledra Advertising (Joined Cases C-8/15 P to C-10/15 P), depositors of two large Cypriot banks brought actions against the European Commission and the European Central Bank for the partial annulment of the Memorandum of Understanding of 26 April 2013 adopted jointly by the ESM and the Republic of Cyprus and also for compensation for damages allegedly suffered following the request for financial assistance and the ensuing restructuring of the two banks in question.

The ECJ had already been called upon to rule on judicial protection questions raised by the ESM framework. Created in order to provide, where needed, financial assistance to the Member States whose currency is the euro, the ESM was instituted through an international agreement between euro area Member States – the Treaty establishing the ESM, concluded in Brussels the 2th February 2012, in force since the 27th September 2012. Thus, the ESM Treaty is not part of the EU legal order, as confirmed by the ECJ in the famous Pringle judgment (C-370/12). As a consequence, when creating the ESM, or acting within its framework, Member States do not act within the scope of application of EU law for the purposes, in particular, of Article 51(1) CFREU. Individuals seeking to challenge Member States’ measures adopted pursuant the conditions laid down in a Memorandum of Understanding would not, therefore, find in the preliminary ruling mechanism an indirect means of access to the ECJ in order to assess their compliance with EU law and, in particular, the CFREU as the former was not in question and the latter was hence out of reach.

What the above mentioned judgments, and especially Ledra Advertising, emphasize is the link nonetheless existing between the ESM framework and the EU legal order. Quoting Alicia Hinarejos (EU Law Analysis), in order to carry out its functions, the ESM “borrows” two EU institutions, the European Commission and the European Central Bank, two thirds of the infamously known Troika. The question is whether (and, if so, when) EU institutions’ actions within the ESM framework might be reviewed and, when harmful, give rise to compensation under EU law and, in particular, in light of the CFREU.

Continue reading “Editorial of October 2016”

Summary of Brasserie du Pêcheur & Factortame – C-46/93 and C-48/93


by José Ricardo Sousa, student of the Master's degree in EU Law of UMinho

Keywords: liability of the state; legislator; claims; repair; individual’s rights

Court: CJEU | DateMarch 5th 1996 | Cases: C46/93 and C-48/93 | Applicants: Brasserie du Pêcheur vs Federal Republic of Germany

Summary: This judgment contains two similar cases connected to the same matter: liability of the State.

In the first case, the French company Brasserie du Pêcheur was obliged to cease their exportations from Germany due to German authorithies’ allegations that the beer did not fulfill purity requirement. European Comission interfered in this case and stated that this provisions were contrary to article 30 of EEC Treaty and brought an infringement proceedings against German Federal Republic. On 12th March 1987, the court confirmed EC’s arguments and consequently condemned the German act. Therefore, Brasserie du Pêcheur moves another action to reclaim their losses. The Court had doubts related to the limits of liability of the State and internal law and so they decided to send a question to the CJEU.

In the second case, Factortame intented an action in High Court of Justice with the purpose to challenge the compatibility of Part II of the Merchant Shipping Act with article 52 of the EEC Treaty. This law predicted a new register for British fishing boats and it pretended to obligate vessel’s registration, including those already registered, according to some conditions relating to nationality. The boats that couldn’t be registered were forbidden to fish. In another previous judgement, CJEU considered that this law was contrary to Communitary law, but it was not contrary that all the boats in UK suffered more controled by the authorities. On 4th August 1989, European Comission brought infringement proceedings against UK to suspend nationality requests because they were contrary to articles 7, 52 and 221 of the EEC Treaty. Afterwards, the Court decided to call the intervenients to show the amount of claims, however the Court had doubts in what refers to include a claim for inconstitutional behaviour and send a question to CJEU.

Continue reading “Summary of Brasserie du Pêcheur & Factortame – C-46/93 and C-48/93”

Summary of Francovich – 6/90

by José Ricardo Sousa, student of the Master's degree in EU Law of UMinho

Keywords: social policy; liability; directive implementation; failure to fulfil an obligation; compensation.

Court: CJEU | DateNov. 19th 1991 | Case: 6/90 | Applicants: Andrea Francovich vs Italian Republic

Summary: The Directive 80/897 goal was to assure a minimum protection for all European workers in case of bankruptcy of a company. For this purpose, it predicted specific guarantees for the payment of claims relating to debt remuneration. Italian Government didn’t implement the mentioned policy in time. Mr Francovich and Mrs Bonifaci filed in court arguing that it was the Italian Government’s obligation to implement the Directive 80/897 and so they claimed a state compensation. The national court suspended the case and referred the following questions to CJEU:

“Under the system of Community law in force, is a private individual who has been adversely affected by the failure of a Member State to implement Directive 80/897 — a failure confirmed by a judgment of the Court of Justice — entitled to require the State itself to give effect to those provisions of that directive which are sufficiently precise and unconditional, by directly invoking the Community legislation against the Member State in default so as to obtain the guarantees which that State itself should have provided and in any event to claim reparation of the loss and damage sustained in relation to provisions to which that right does not apply?”

Continue reading “Summary of Francovich – 6/90”