Mapping the Epistemic Gap: A Systematic Evidence Map of Supranational AI Regulation and Innovation Outcomes, 2018–2025

This article identifies a pronounced 2-to-19 asymmetry in the emerging literature on supranational AI regulation and innovation outcomes: only one included study employs a quasi-experimental quantitative design, while one additional study provides case-based empirical evidence directly linking regulation to an innovation outcome; the remaining 19 studies are assigned to the mechanism-oriented stream, including empirical or mixed-methods studies that do not directly estimate the regulation–innovation relationship. Both Stream A studies concern GDPR-related mechanisms, and no included study quantitatively estimates the innovation effects of the EU AI Act. Using PRISMA-style reporting, the article presents a systematic evidence map of literature from 2018–2025 across venture capital, startups, patents, and talent mobility. The evidence-map design makes the search, screening, and coding process transparent while acknowledging that the field is too recent and heterogeneous for pooled causal synthesis. Mixed effects are the modal directional coding (56%). The central finding is therefore not that supranational AI regulation has a single positive or negative innovation effect, but that mechanism claims currently outpace causal empirical evidence.