A UK-US-EU research team is using AI chatbots to simulate how tech companies like Google or Amazon would react to different competition laws, before those laws are ever passed. Current competition law is reactive—regulators can only act after a company has already harmed consumers or rivals. This project addresses a fundamental gap: regulators have no reliable way to predict how a new rule will actually behave when faced with profit-driven, adaptive companies that can restructure or relocate overnight. The researchers build simulated market environments where large language models act as firms, consumers, and regulators, then test different legal designs to see which ones work and which backfire. If successful, this *in silico* methodology could transform how governments design regulation for digital markets. Instead of guessing whether a new antitrust rule will curb monopoly power or simply push it elsewhere, policymakers could run simulations first. The approach is not limited to competition law—it could be applied to any legal field where institutional design matters. This is fundamental methodological research: it asks whether computational models can do causal legal reasoning at all, and if so, how.
View original technical description
This collaborative research project pioneers a novel methodology in comparative legal studies: in silico legal comparison. By integrating insights from empirical comparative law and experimental legal studies, it advances an emerging field that leverages Large Language Models (LLMs) for comparative institutional and legal analysis. The field of competition law is particularly fruitful for such research, in view of the different substantive law approaches followed in each jurisdiction, the variety of institutional designs and the commonality of the competition problems and business conducts examined, given the global strategies of digital market players and the influence of economic analysis in competition law. The research will make the following three contributions: First, the research examines how and why the United States (US), the United Kingdom (UK), and the European Union (EU) differ in their regulatory responses to addressing competitive risks generated by AI technologies and digital ecosystems in their respective markets. It maps institutional frameworks and substantive rules across these three jurisdictions, identifying key differences in regulatory/antitrust approaches. It then employs AI simulation tools to model how businesses respond to the different components of these regulatory environments. By testing different scenarios using vignettes, we examine how competition law enforcement and digital competition regulation can address competitive risks related to AI and digital ecosystems. These simulations reveal how tech companies can adapt their strategies in response to regulatory pressure and how consumers and other stakeholders may react to them. We demonstrate precisely which regulatory designs are likely to be effective in specific legal contexts - and which might backfire. The research thus comprehensively examines the institutional frameworks and substantive regulations governing AI- and digital ecosystem-related competition risks, illuminating significant differences in antitrust enforcement approaches and, crucially, how such antitrust regimes function when confronted with adaptive market participants. Second, the research aims to uncover the fundamental philosophical principles and societal values shaping each legal system's distinct strategy towards AI and digital ecosystems generated competition risks, proceeding to a 'law in action' analysis in an artificial (in silico) environment. Finally, the research bridges legal scholarship and computational studies, thus developing a methodology with the potential to transform comparative legal analysis. This new approach to legal comparison requires exploring the causal capabilities of LL.Ms for comparative legal research purposes and their application to causal reasoning in institutional analysis, while accounting for the different types of causality employed in law and data science. This research thus establishes a foundation for future comparative legal and empirical research, not only in competition law but also across other legal fields, utilising the in silico legal comparison methodology.
Plain English summaries and category classifications on this site are generated by AI and may not perfectly reflect the original research.
Is something wrong? Let us know