Justifying AI regulation: Examining multi-stakeholder responses to the AI Act

The regulation of artificial intelligence (AI) has sparked discussions among stakeholders since the introduction of the European AI Act (AIA) regulatory proposal. This study examines how stakeholders justified their arguments around AI and its regulation in multi-stakeholder feedback on the first draft of the AIA. Drawing on the framework of social justifications (Boltanski and Thévenot, 2006) and leveraging Natural Language Inference (NLI) for assisted reading, the paper identifies areas of contestation and similarity between stakeholders from the technology industry, academia, non-governmental organizations, and the public sector. The findings show that stakeholders employed similar justifications to differing ends, suggesting that there exists some shared conception or tactical recognition of the types of justifications that are considered effective in this context. The paper argues that industry-driven discourse influences how stakeholders argue for and against regulating AI, and that justifications differ in the means they propose for achieving a similar end goal.