Lex Imperfecta: Leveraging (Un)supervised Machine-Learning Tools for Optimizing Digital Markets Regulations

Research in the emerging field of law and computer science is challenging our understanding of how firms comply with laws in the digital economy and whether the traditional deterrence-based approach to regulation is still effective. Studies using computational tools have shown that regulations, such as the EU GDPR, are often ignored by online firms, with high levels of non-compliance with EU privacy regulations. Additionally, there is increasing evidence of low compliance by online firms with consumer protection laws. In a digital society, such widespread non-compliance can lead to significant societal costs. Given this, there is an urgent need to re-evaluate the current deterrence-based regulatory approach and assess whether it remains suitable for the digital economy. This proposal aims to conduct such an evaluation using new tools and an interdisciplinary approach. First, we will develop and use computational tools to collect empirical data on compliance with privacy, consumer, and competition laws in EU Member States over time. This will enable us to create the most comprehensive database on compliance, including detailed information at the country, industry, and firm levels. In the second stage, we will use this database to critically evaluate the theories of deterrence and enforcement that underpin the EU’s regulatory approach, considering the unique features of digital markets—such as scale, speed, and technological sophistication. The final stage will focus on designing laws that support automated enforcement mechanisms. We will explore what makes a law suitable for automation, including the terms, definitions, and procedures that could enhance scalability and improve enforcement, thereby making it a more effective deterrent.