Revista Ítalo-Española de Derecho Procesal
pp. 133-142
Madrid, 2025
DOI: 10.37417/rivitsproc/2911
Marcial Pons Ediciones Jurídicas y Sociales
© Maria Dymitruk
ISSN: 2605-5244
Recibido: 22/11/2024 | Aceptado: 21/02/2025
Editado bajo licencia Creative Commons Attribution 4.0 International License.
AI-Based Automation of the European Small Claims Procedure:Legal Framework Mapping
Maria Dymitruk*
University of Wrocław
ABSTRACT: The paper explores current legal regulations regarding AI-based automation (full or partial) of the European Small Claims Procedure (ESCP). It analyzes European-level legislation, focusing on personal data protection (specifically Article 22 of the GDPR) and human rights (mainly Article 47 of the Charter of Fundamental Rights), while also mentioning the possibility of establishing a legal basis for ESCP automation in the national legal orders of EU Member States’ legal systems.
KEYWORDS: European Small Claims Procedure; ESCP; automation; AI; artificial intelligence.
SUMARIO: 1. INTRODUCTION.— 2. LITERATURE REVIEW: THIS IS CERTAINLY NOT THE FIRST PAPER ON THE USE OF AI IN ESCP.— 3. ESCP AUTOMATION MODELS AND WHY DISTINGUISH BETWEEN THEM?.— 4. LEGAL FRAMEWORK FOR ESCP AUTOMATION.— 5. REGULATION ESTABLISHING A EUROPEAN SMALL CLAIMS PROCEDURE.— 6. HUMAN RIGHTS REGULATION.— 7. GENERAL DATA PROTECTION REGULATION.— 8 NATIONAL REGULATIONS.— 9. CONCLUSION.— REFERENCES.
There is perhaps no more pertinent topic in contemporary discourse within the realm of law and technology than the utilization of artificial intelligence (AI) advancements for automation within the legal system. Of particular interest is the prospect of employing AI to automate processes within the judiciary, addressing the perennial challenge of insufficient resources and case backlogs across jurisdictions.
Such AI-driven automation can manifest in a myriad of forms. These range from simple tools that streamline basic clerical tasks in court secretariats. They extend to algorithms supporting ancillary functions within the judiciary, such as anonymizing court decisions for public dissemination. Such ancillary tasks are important in terms of supplementing the intellectual work carried out within the judiciary. However, they do not fall within the scope reserved for human judges and the processes of law application. This third form of automation precisely concerns substantive decision-making processes. It augments or even replaces human involvement in the adjudicative process, ultimately culminating in court verdicts. However, even the automation of substantive tasks in the application of law by courts is not uniform and encompasses a variety of automation models. These models can differ from each other primarily in terms of the level of involvement of the human decision-maker in the final outcome of the law application process. Additionally, differences may arise from the timing of the AI system’s intervention in the human decision-making process and the role it plays in such intervention. Further elaboration on this subject will be provided later in the paper.
The paper will not address AI in the judiciary in general but will focus on one specific type of procedure: the European Small Claims Procedure (ESCP). Various forms of substantive automation—specifically, the automation of processes aimed at resolving civil or commercial cases—will be analyzed. The initial step towards determining the scope of potential implementation activities involves outlining the legal framework in which such automation takes place.
The ESCP is an unusual judicial procedure: it is optional, written, characterized by a high degree of formalization, and operates uniformly across the European Union (EU), making it multilingual. This makes it a valuable subject for analysis regarding the feasibility of court automation. This is an area where not all theories about the use of AI in courts are applicable. These differences support, rather than undermine, the AI-based automation mentioned in the paper’s title. It is therefore not surprising that some researchers have already addressed this topic.
Veersalu and Hoffmann advocated for the establishment of an EU-wide online ESCP platform, managed by the European Commission. This platform, utilizing AI and electronic communication between the court and the parties, would streamline procedures and promote practical utilization of ESCP (Veersalu, Hoffmann, 2023). However, the authors emphasized that the prospect of automating adjudication within the ESCP does not appear realistic. Withal, Simaitis, Vėbraitė, and Markevičiūtė pointed out that the potential for AI-based ESCP automation extends beyond automated translations. In their opinion, it encompasses addressing recurring ESCP issues and providing assistance to parties and judges (Simaitis, Vėbraitė, Markevičiūtė, 2022, p. 133). Furthermore, Veersalu and Hoffmann referenced several legal acts pertinent to defining the legal framework for ESCP automation, including the then-planned Regulation of the European Parliament and of the Council laying down harmonised rules on artificial intelligence and amending certain union legislative acts (AI Act), the Regulation (EU) 2016/679 of the European Parliament and of the Council of 27 April 2016 on the protection of natural persons with regard to the processing of personal data and on the free movement of such data, and repealing Directive 95/46/EC (GDPR), and regulations concerning the right to a fair trial and the prohibition of discrimination (Veersalu, Hoffmann, 2023, p. 159).
Taking into account the aforementioned scientific publications, what does this paper aim to contribute to the existing body of work? It seeks to integrate discussions of AI applications within the ESCP by exploring the specific regulations governing potential deployment activities. Furthermore, it aims to ground theoretical considerations about potential AI usage within the ESCP in the context of current legal regulations
The use of AI to automate court decisions can be categorized into two basic models: the full court automation model or the semi-automation model. The space for fully automating judges’ tasks—replacing the result of an AI system with a court judgment, which is binding on the parties to the proceedings and permanently affects their legal situation—is limited and subject to several additional caveats, but it is not entirely excluded (Dymitruk, 2023). Notably, under the rules governing EU cross-border procedures, Article 8 of Regulation (EC) No 1896/2006 of the European Parliament and of the Council of 12 December 2006 creating a European order for payment procedure, allows for automation. This provision enables the automated examination of the prerequisites outlined in Articles 2 to 7 of the Regulation and the validation of the claim’s legitimacy, marking a significant incursion of AI into a decision-making area traditionally occupied by human decision-makers (Kościółek, Banaszewska, 2016). This example illustrates that full judicial automation can apply to scenarios where an AI system issues a judgment, as well as cases where the system is authorized to make relevant procedural decisions preceding the final judgment within a given court instance.
The potential delegation of the power to make a binding determination, even if only a fragment, of a court case into the hands of AI-powered computer software is not the sole form of automation within the judiciary. Generally less controversial are semi-automation models, which do not fundamentally alter the existing decision-making paradigm within the judiciary, allowing the final decision on the content of a judgment to remain in the hands (or, if you prefer, the mind) of a human judge. In semi-automation constructs, AI tools serve in a subservient role to the judge, who retains the sole authority to decide the fate of judicial proceedings and their parties. However, this supportive role of AI can manifest in different ways, and the AI system’s intervention in the judge’s autonomous thought process can occur at various stages. Therefore, a distinction should be made between two forms of semi-automation: the first being where the AI system’s support precedes the moment of decision-making by the human judge, and the second being where AI assistance comes after the judge has made their decision. In the first submodel of semi-automation, the role of the AI system is mainly to ‘hint’ to the judge the decision to apply the law (with the judge having the discretion to agree or disagree), while in the second submodel, its role is to evaluate the judge’s reasoning and potentially suggest improvements (such as highlighting overlooked case law or evidence, or pointing out contradictions in the evidence).
Phenomena inherent in contemporary AI systems, such as the lack of explainability, along with issues related to their utilization in decision-making processes—particularly the automation bias (the cognitive error in relation to AI systems, which are formally intended only to provide advice, but may exert an excessive persuasiveness)—mean that semi-automation models, where the AI system’s support precedes independent decision-making by a human, pose the risk of ‘quasi-automation’ (MSI-NET, 2018). This entails a discrepancy between the formally established automation model (in this case, suggestion-based semi-automation) and its operational practice (which de facto results in full automation under the guise of semi-automation).
Why does this distinction between automation models matter for the purposes of this paper? It serves to elucidate (and later delve deeper into) the non-uniformity in the potential application of AI in the ESCP, and to illustrate the variances among different automation models within the legal framework they are intended to operate under—de lege lata—in the European Union.
The use of artificial intelligence to automate the judicial process of applying the law (even only to support human judges) requires not only the creation of a properly functioning AI system capable of conducting the multitude of actions and reasoning involved in adjudication, but also a thorough analysis of the admissibility of using AI tools in such a context. It is challenging to imagine introducing artificial intelligence into the legal sphere without verifying whether such use would contradict the existing legal framework. In the case of automation of the ESCP, the legal framework would need to be embedded in two levels of regulation: EU regulation and national regulation.
At the EU level, the legal framework would be shaped by procedural regulations, including, in particular, the provisions of Regulation (EC) No 861/2007 of the European Parliament and of the Council of 11 July 2007 establishing a European Small Claims Procedure, as well as regulations concerning personal data protection and, more broadly, regulations concerning human rights related to the interface between automation and court procedures.
At the national level, the legal framework would be shaped by general procedural regulations (applicable in civil and commercial cases), potentially constitutional regulations (especially regarding the scope determining the possible use of AI within the judiciary), possible regulations on personal data protection at the member state level detailing the GDPR, as well as any other normative acts addressing the issue of automation of decision-making processes using AI.
The ESCP regulation itself does not mention in any way the possibility of automating the decision-making processes taking place within the ESCP. Thus, it does not introduce the possibility of either automating the formal review of the lawsuit or deciding on the merits of the case. As already indicated, this is done differently by the Regulation creating a European order for payment procedure, which in Article 8 introduces the basis for conducting an automated examination of the claim. Kościółek and Banaszewska, analyzing whether such a wording of the provision makes it possible to recognize the EU legislature’s consent to the issuance of a European order for payment by AI, conclude that the wording of Article 8 does not exclude the possibility of applying an automated examination of the claim both in terms of formal conditions and substantive assessment of the merits of the claim (Kościółek, Banaszewska, 2016). Indeed, the authors point out that the provision explicitly speaks of two levels of examination of the claim, and then allows for the possibility of applying an automated examination procedure, without stipulating that it should apply only to one of the areas mentioned therein (Kościółek, Banaszewska, 2016, p. 33).
Vytautas argues differently, stating that ‘it is obvious that the computer is not capable of checking either the sufficiency of the evidence, its validity, or the validity of the claim itself,’ hence the author limits the scope for automation only to the system’s examination of whether Form A has been filled out correctly (Vytautas, 2011). The author supports this conclusion with the wording of recital 11 to the Regulation, which stipulates that the procedure should be based, to the largest extent possible, on the use of standard forms in any communication between the court and the parties to facilitate its administration and enable the use of automatic data processing. For a full evaluation of Vytautas’ position, it should be taken into account that his opinion was expressed in a different technological landscape than today. The scale of AI development over the past 10 years allows to draw bolder conclusions about AI’s application potential today.
Regardless of one’s assessment of the extent to which the Regulation establishing the European order for payment allows for automation, it is undeniable that a similar provision is absent in the ESCP Regulation. This implies that the issues examined in the paper are not directly addressed in this legislation—despite its crucial role in shaping the ESCP, but it is not the sole regulatory instrument at play. Therefore, the regulatory framework for potential AI-based automation (full or partial) must be sought elsewhere.
One of the main objections to both the model of full automation and semi-automation, shaped in a way that does not prevent quasi-automation, remains the failure to fulfill all the requirements of a fair trial—a concept at the core of law-abiding performance within the justice system of the European Union (Dymitruk, 2023). The EU’s treaty-level regulation, specifically Article 47 of the Charter of Fundamental Rights, plays a key role in this analysis.
It should be emphasized that the concept of a fair trial is not unique to the EU legal order. It finds its source in the legal framework established through the activities of the Council of Europe (as seen in Article 6 of the European Convention on Human Rights and the jurisprudence of the European Court of Human Rights in Strasbourg derived from it), global international agreements (including Article 14 of the International Covenant on Civil and Political Rights), and often in national laws (of constitutional or procedural nature).
The elements of the fair trial concept that pose a particularly challenging relationship with ESCP automation include judicial independence, transparency, and the right to be heard. Conversely, the requirements of certainty and efficiency (especially speed) are fulfilled in an exemplary manner, sometimes even better than in cases where justice is solely administered by human hands (Dymitruk, 2023). Additionally, distinctive features of the ESCP, such as its non-coercive nature and predominantly written format, serve as facilitators in this context. The voluntary nature of the ESCP, in comparison with mandatory national procedures, is likely to mitigate issues with fully realizing the fair trial. If parties to the proceedings consent to the automation (whether partial or full) of the procedure, then the challenges in implementing fair trial requirements may become at least partially obsolete. Indeed, parties may prioritize obtaining a speedy resolution over complete transparency, particularly regarding the necessity of providing explanations for legal decisions. Positioning the ESCP as a kind of automated “zero instance” could significantly enhance the attractiveness of the procedure compared to other institutional solutions, effectively addressing one of its primary practical challenges—its low utilization in practice.
Equally significant within the realm of human rights regulation for the automation of ESCP would also be:
— the prohibition of discrimination (Article 21 of the Charter of Fundamental Rights), particularly concerning algorithmic bias and resulting possible discrimination based on race, gender, or other factors;
— the right to respect for private life (Article 7 of the Charter of Fundamental Rights), which, in the context of algorithmic decision-making, has led to court rulings (e.g., Rechtbank Den Haag ruling of March 6, 2020, ECLI:NL:RBDHA:2020:865);
— the right to the protection of personal data (Article 8 of the Charter of Fundamental Rights), as developed by GDPR regulations, which will be analyzed in more detail in the next paragraph.
The GDPR does not explicitly mention artificial intelligence, yet it is of considerable importance in the context of ESCP automation. It’s challenging to conceive of any automated processing of data within claim files that wouldn’t involve personal data (pertaining to parties to the proceedings, witnesses, or any other natural person). According to recital 20 of the GDPR, the regulation is fully applicable to the activities of courts and other judicial authorities, providing the regulatory framework for ESCP automation. This includes principles of personal data processing, the legal basis for personal data processing, requirements for transparency and the rights of data subjects, as well as the crucial issue of automated decision-making in individual cases.
From the perspective of automating the ESCP, Articles 22, 13, 14, and incidentally Article 9 of the GDPR, remain relevant, all within the context of automated processing of personal data. Article 22 plays a central role here, establishing the conditions for establishing a potential legal basis for ESCP automation based on personal data. Indeed, it not only introduces a general right not to be subject to a decision based solely on automated processing, including profiling, which produces legal effects concerning him or her or similarly significantly affects him or her (Article 22(1) GDPR), but also introduces exceptions to such a generally formulated prohibition (Article 22(2) GDPR). The analysis of these exceptions leads to the conclusion that the only potentially implementable legal basis for the automation of judicial decisions in practice remains the authorization contained in EU or Member State law to which the controller (in this case, the court) is subject (Article 22(2)(b) GDPR) (Almada, Dymitruk, 2022). This means that, in order to establish a workable ESCP automation model, the relevant legal basis should be found in a norm of EU law (preferably in the provisions of the ESCP regulation itself), which would require an amendment of the regulation. An alternative legal basis in national laws could also provide an effective foundation for automation but would result in further fragmentation of ESCP regulation between different member states, which is not conducive to the harmonization of ESCP standards across the EU.
Article 22 of the GDPR applies only if the decision is based solely on automated processing. What does this mean in the context of the automation models highlighted in the paper? The full automation model falls within the scope of Article 22, as it is clearly based solely on automated processing of personal data and produces legal effects on the data subject. An analysis of the Article 29 Data Protection Working Party’s position on defining ‘solely automated decision-making’ using the concept of meaningful human involvement (Article 29 Data Protection Working Party, 2018), along with an awareness of the risk of ‘quasi-automation’, suggests that it is not excluded to determine the applicability of Article 22 of the GDPR to a semi-automation model, which would provide suggestions to the judge in the ESCP. From the perspective of the two distinguished sub-models of semi-automation, only the model evaluating the decision of the human judge is completely outside the scope of application of Article 22(1) GDPR in this context.
At the same time, if the applicability of Article 22 GDPR to ESCP automation is established, it remains the duty of the court as the data controller to implement suitable measures to safeguard the data subject’s rights, freedoms, and legitimate interests. It is the responsibility of the legislator (whether EU or national) who decides to introduce an appropriate legal basis for the automation of the ESCP to define the catalogue of these safeguards. However, the catalogue of safeguards listed in Article 22(3) of the GDPR does not apply in this case, as it is mandatory only if the other two grounds for automatic data processing (i.e., Article 22(2)(a) and (c)) apply. Nevertheless, this catalogue represents an absolute minimum rather than an excessive requirement, so safeguards covering the right to obtain human intervention from the controller, to express one’s point of view, and to contest the decision should certainly apply here. These safeguards also coincide with the requirements of a fair trial, forming a complementary regulation with the procedural requirements of a fair trial in the context of ESCP automation.
The automation of ESCP based on special categories of data within the meaning of Article 9 of the GDPR would face even further restriction. According to Article 22(4) GDPR, automated decisions cannot be based on special categories of personal data unless point (a) or (g) of Article 9(2) applies and suitable measures to safeguard the data subject’s rights, freedoms, and legitimate interests are in place. In the context of ESCP automation using so-called sensitive data, this means in practice that the premise of ‘substantial public interest’ must be fulfilled (Almada, Dymitruk, 2022).
In addition to the regulatory core expressed in Article 22 of the GDPR, Articles 13(2)(f) and 14(2)(g) of the GDPR are relevant in the context of ESCP automation. They create an obligation on the part of the controller to provide the data subject with information about the existence of automated decision-making and, at least in those cases, meaningful information about the logic involved, as well as the significance and the envisaged consequences of such processing for the data subject (for further discussion on the existence of a right to explanation under the GDPR, see Goodman, Flaxman, 2017; Wachter, Mittelstadt, Floridi, 2017).
Determining the legal framework for potential automation of the ESCP requires relevant analyses of the legal orders of the member states. The ESCP is a procedure carried out by national courts, and their autonomy in this process is preserved. Therefore, any analysis that omits aspects of national procedural law, and often constitutional law as well (e.g., regarding the admissibility of automation of judicial tasks), would be incomplete. However, an analysis of the legal systems of all twenty-seven member states in this regard goes significantly beyond the scope of this paper.
It is also important to note that national legal orders may include regulations other than procedural regulations that are relevant for determining the legal framework for the automation of ESCP. These may include legislation detailing the GDPR or even introducing a specific legal basis for automated decision-making (see more Malgieri, 2019), or even entirely separate legislation, including laws typically dedicated to the judiciary. Given the significance of issues related to algorithmic decision-making and the increasing interest of national legislators in artificial intelligence topics, a growing body of legislation in this area can potentially be expected.
De lege lata, the regulatory framework for potential automation of the ESCP remains fragmented across various pieces of legislation, both EU and national. Navigating and assembling the full legislative puzzle of potential implementations is no easy task. The situation has not changed with the final adoption and publication of the AI Act. This regulation has become an additional important landmark (if not the most important) in the broad regulatory landscape of AI-based automation, including judicial automation, especially given the inclusion of AI systems used in the judiciary in the high-risk areas to which the vast majority of AI Act provisions are addressed. This will certainly transform the regulatory framework for ESCP automation, but it will not replace the area already described in this paper. The conclusions drawn from it will not become obsolete; rather, new areas of legal analysis will be added, raising questions about the interface of individual provisions of the AI Act with the regulations discussed (for example, the relationship of Article 86 of the AI Act to the provisions of the GDPR).
However, undoubtedly, the most important conclusion from the analyses carried out here remains the need to update the ESCP regulation. This update is necessary not only to introduce a legal basis for automation but also to incorporate appropriate procedural guarantees for litigants (and possibly other data subjects involved in the procedure). These guarantees will be a consequence and a blend of data protection regulation, procedural regulation and AI management requirements, often intersecting at both the EU and national levels.
Almada, M., & Dymitruk, M. (2022). Data protection and judicial automation. In Research Handbook on EU Data Protection Law. Edward Elgar Publishing, 270-289.
Article 29 Data Protection Working Party. (2018). Guidelines on Automated Individual Decision-Making and Profiling for the Purposes of Regulation 2016/679.
Committee of Experts on Internet Intermediaries (MSI-NET). (2018). Algorithms and Human Rights – Study on the human rights dimensions of automated data processing techniques (in particular algorithms) and possible regulatory implications.
Dymitruk, M. (2023). Wykorzystanie sztucznej inteligencji w postępowaniu cywilnym (doctoral dissertation). University of Wrocław.
Goodman, B., & Flaxman, S. (2017). European Union regulations on algorithmic decision-making and a “right to explanation”. AI magazine, 38(3), 50-57.
Kościółek, A., & Banaszewska, A. (2016). Automatyzacja czynności orzekania w europejskim postępowaniu nakazowym. E-Wymiar sprawiedliwości w aspekcie europejskim, Poznań 2016, 37-52.
Malgieri, G. (2019). Automated decision-making in the EU Member States: The right to explanation and other “suitable safeguards” in the national legislations. Computer law & security review, 35(5), 105327.
Nekrošius, V. (2011). Postępowanie w sprawie europejskiego nakazu zapłaty, Aurea praxis, Aurea theoria: Księga Pamiątkowa ku czci Profesora Tadeusza Erecińskiego, Warszawa, LexisNexis, 1213-1231.
Simaitis, R., Vėbraitė, V., & Markevičiūtė, M. (2022). European Small Claims Procedure in the realm of the other European proceedings. Revista Ítalo-española de Derecho procesal, (1), 123-136.
Veersalu, K., & Hoffmann, T. (2023). Automation Within a Novel Platform for the European Small Claims Procedure. TalTech Journal of European Studies, 13(2), 152-176.
Wachter, S., Mittelstadt, B., & Floridi, L. (2017). Why a right to explanation of automated decision-making does not exist in the general data protection regulation. International data privacy law, 7(2), 76-99.
* Faculty of Law, Administration and Economics, University of Wrocław. https://orcid.org/0000-0003-1003-9083