The dissemination of data and the circulation of scientists are widely regarded as pivotal factors in the advancement of science. However, several countries, including the European Union, Canada and the United States, have recently adopted restrictive policies designed to protect scientific results against the risks of economic espionage, hostile military use or terrorist diversion. This article examines the legal frameworks that have been implemented in France to protect the nation's 'scientific and technological capability'. The French approach encompasses the establishment of restricted zones (ZRR), comprising three constituent elements: the creation of protected areas, the restriction of access to these areas, and the limitation of data circulation. This has resulted in a fragmentation of scientific communities, based on ‘communities of trust’ that are seeking to restore conditions favourable to the production of scientific knowledge in restricted and controlled areas. The article illustrates the resurgence of a legal framework for scientific research centred on national interests, suggesting a possible retreat in the openness of science.
The Chinese AI regulatory model has been described as state-driven, where the Chinese government seeks technological supremacy by deploying AI to entrench power through surveillance, control, and cooptation of public and private sectors. It is portrayed as a comprehensive, top-down approach with little room for agency, where autocracy and AI innovation mutually reinforce each other. However, this paper provides a more nuanced picture of China's AI governance. We argue that the Chinese government, at both central and local levels, has adopted a subtle and collaborative approach that integrates state action with private capabilities in dynamic ways that are not necessarily autocratic in nature. We demonstrate this through a framework analyzing the Chinese state's role as investor, regulator, and procurer in relation to private AI companies. First, as an investor, instead of being a controlling shareholder, we find that the Chinese government (or its entities) is a non-controlling shareholder who exerts indirect influence through overlooked mechanisms, while preserving governance powers for corporate boards. Second, as a regulator, contrary to assumptions that regulation is enforced through coercive command-and-control methods, we find that local governments rely on financial support to incentivize AI firms to align with the central government's objectives. Finally, as a procurer, contrary to the notion that autocracy reinforces innovation through unrestricted data sharing, we show that legal and institutional restrictions apply, and AI firms often have incentives to use non-government data. These findings complexify conventional narratives of Chinese AI governance and offer insights into the evolving state-private relationship, with implications for global AI governance debates.
