Data provision as an actionable subsidy: examining the de facto specificity of national data-sharing policies for Generative AI under the SCM Agreement
Journal of International Dispute Settlement · 13.01.2026 · 17 (1)
DOI: 10.1093/jnlids/idag006 ↗Kurz gefasst
Fachartikel in der Zeitschrift Journal of International Dispute Settlement, online erschienen am 22.05.2026. Verfasst von Zihan Chen.
Abstract (Originalsprache)
The rise of Generative AI (GAI) has made state-facilitated ‘data pools’ a cornerstone of national industrial policy. This article examines whether these policies constitute actionable subsidies under the World Trade Organization SCM Agreement (Agreement on Subsidies and Countervailing Measures). It argues that while such programmes appear facially neutral, they exhibit de facto specificity under Article 2.1(c) due to the GAI industry’s unique technical barriers. To address the jurisdictional gap in service subsidies, this article proposes a ‘bridging interpretation’ that characterizes GAI models as digital goods. By distinguishing genuine open data from targeted support and exploring legal safe harbours like DEPA, this article concludes that the SCM framework is essential to preventing protectionist fragmentation and ensuring fair competition in the global AI economy.
Zitieren
APA
Chen, Z. (2026). Data provision as an actionable subsidy: examining the de facto specificity of national data-sharing policies for Generative AI under the SCM Agreement. Journal of International Dispute Settlement, 17(1). https://doi.org/10.1093/jnlids/idag006
Vancouver
Chen Z. Data provision as an actionable subsidy: examining the de facto specificity of national data-sharing policies for Generative AI under the SCM Agreement. Journal of International Dispute Settlement. 2026;17(1). doi:10.1093/jnlids/idag006
BibTeX
@article{Chen2026,
author = {Chen, Zihan},
title = {Data provision as an actionable subsidy: examining the de facto specificity of national data-sharing policies for Generative AI under the SCM Agreement},
journal = {Journal of International Dispute Settlement},
year = {2026},
volume = {17},
number = {1},
doi = {10.1093/jnlids/idag006}
}
EUVIMED