The data sovereignty framework has emerged across multiple jurisdictions through different mechanisms. The European Union's GDPR established privacy-based data protections that have de facto limited extraction. India's data localization requirements assert national authority over data generated within Indian territory. China's comprehensive data governance regime treats data as a strategic national asset. Multiple Global South nations have raised data sovereignty concerns in international AI governance discussions.
The Chang reading of data sovereignty connects it to the broader history of developmental sovereignty. The pattern is recurrent: a nation generates valuable resources (raw materials in earlier periods, data in the contemporary period); foreign actors claim access to those resources on terms favorable to themselves; the nation asserts sovereign rights to shape the terms of access; the international system pressures the nation to abandon the assertion in favor of 'efficient' market arrangements that systematically transfer value to the foreign actors. The pattern that Chang has documented for textiles, semiconductors, and pharmaceuticals now recurs for data.
The relevance to the AI training data question is direct. The current legal framework, in which AI companies treat globally scraped data as freely appropriable for commercial training, represents a contemporary version of the colonial extraction practices that successful developers historically resisted. Data sovereignty arguments provide the framework through which contemporary nations and communities can assert legitimate claims on the value created from their data — claims that the current arrangement systematically denies.
The political economy of data sovereignty is contested. Foreign AI platforms argue that data localization and sovereignty requirements are 'fragmenting' the internet and reducing efficiency. The argument is structurally identical to the nineteenth-century British argument that German tariff barriers were fragmenting the European market and reducing efficiency. In both cases, the argument serves to delegitimize the policy interventions that developing economies require to assert genuine agency in shaping their participation in the global system.
The phrase 'data sovereignty' came into common usage around 2013–2015, driven initially by post-Snowden concerns about American surveillance access to globally hosted data. The framework expanded subsequently to encompass concerns about commercial data extraction by AI platforms and the cultural and political consequences of dependence on foreign-controlled data infrastructure.
The Chang reading specifically connects contemporary data sovereignty arguments to the longer history of developmental sovereignty struggles. The connection makes visible the continuity between contemporary and historical patterns and provides analytical tools for assessing which contemporary claims are likely to succeed and which are likely to be defeated by the same mechanisms that defeated earlier sovereignty assertions.
Data as strategic resource. Recognition that data is not neutral raw material but the foundation on which AI capabilities are built and through which value is captured.
Sovereignty continuity. The structural identity between contemporary data sovereignty claims and historical developmental sovereignty claims against earlier forms of foreign extraction.
Extraction critique. The current framework of unrestricted commercial data scraping treated as a contemporary version of historical colonial resource extraction.
Policy space contestation. The ongoing struggle over whether nations will retain authority to shape the terms on which data generated within their territories is available for commercial AI training.