Kajava, K., Öhman, E., Takagi, N. M., Nakajima – Wickham, E., & Vitiugin, F. (2025). Post-GPT Policy: Risk and Regulation in EU AI Discourse. Proceedings of the International AAAI Conference on Web and Social Media, 19, 994–1006.
Abstract
The ubiquity of artificial intelligence (AI) systems across society has generated discussion around risks posed by general-purpose AI (GPAI) technologies. At the same time, harmonized AI regulation is being devised in the European Union (EU) to ensure responsible AI development and uptake. This paper presents a mixed-methods study combining approaches from natural language processing (NLP) and qualitative content analysis to study shifting risk discourse in EU policy-related documents and media articles. The research is anchored around two recent events in the AI space: the popularization of GPAI following the release of ChatGPT and the introduction of the EU AI Act (AIA) regulation. The findings reveal a discursive emphasis on cybersecurity, regional imbalance of computational resources, education and AI literacy, and the implications of risks occurring at scale.