Culpa in Code: Exploring Autonomous Algorithmic Collusion under EU Competition Law

Abstract: Algorithms wield an increasingly significant influence in our everyday lives. Both in traditional brick-and-mortar markets, and particularly in the digital markets, companies use algorithms for various operations. Instances of algorithmic collusion have already surfaced. However, a more signifi-cant threat may await. With the increased capabilities of the most modern algorithms, it has been discussed whether algorithms can act collusively independent of any human intention or conduct, in other words, autono-mously. Within the framework of EU competition law, establishing collusion relies on identifying an agreement or, in its absence, a concerted practice. The Court of Justice has adopted a flexible approach to this “criteria of coordi-nation”. However, without giving algorithms a legal personality, one will struggle to argue that autonomous behaviour by algorithms can constitute an agreement. Subsequently, competition watchdogs may face challenges in identifying infringements in such cases. This thesis thoroughly examines the current legislation and jurisprudence related to the “criteria of coordination”, further exploring various academic contributions to determine the meaning and scope. A pivotal aspect ad-dressed is the Court of Justice’s quest to balance the interest of consumers with the freedom of undertakings to intelligently operate on the market, employing various terms and presumptions in this pursuit. Despite the comprehensive exploration, the terms dolus/culpa, denoting fault in legal contexts, are notably absent. This raises questions regarding the role of fault in competition law. Since it is difficult to find fault in a situation involving the collision of autonomous algorithms, this thesis con-templates whether EU competition law necessitates the presence of dolus and/or culpa. In delving into the realm of algorithmic collusion, this thesis strives to apply the current legal framework to a situation of autonomous algorithmic collusion and models in doctrine. Furthermore, a review of contributions of legal scholars to the question is offered. This thesis culminates in an attempt to integrate the current legal frame-work and existing doctrine into a model addressing autonomous algorith-mic collusion. Drawing from the discussion on dolus/culpa and borrowing the tort law concept of duty to care, the presented model arguably aligns with existing jurisprudence while effectively balancing the interests at stake.