Abstract To what extent is AI regulation influenced by frames and discourse coalitions? To address this question, we use complex systems and framing theories to analyse public discourse on facial recognition in the European Union (EU) and the United States. Our discourse network analysis of statements between 2000 and 2022 shows that facial recognition has been framed as a legitimate solution for security issues by governments particularly after 9/11. However, these earlier frames have been increasingly contested, and the dominant discourse shifted from security to ethical concerns. As facial recognition became more intrusive, we observe a diffusion of human rights and privacy frames and an associated gradual desecuritisation of the debate in both cases. We argue that desecuritisation and frame diffusion structured which forms of regulation were perceived as feasible, necessary and legitimate, thereby challenging certain applications of AI surveillance. We also show that this shift in the discursive environment influenced the scope and ethos of US state‐level facial recognition policies and EU legislation such as the General Data Protection Regulation (GDPR) and the AI Act.
Öge et al. (Mon,) studied this question.