5 July 2026

Databases Centralized for AI Development

The Jamestown Foundation  |  Samantha Hoffman

The People's Republic of China (PRC) initiated a new long-term plan to construct sector-specific databases for artificial intelligence (AI) development, announced at the Ninth Digital China Construction Summit on April 29, 2026. The National Data Administration's (NDA) Implementation Plan, issued June 3, 2026, mandates "physically distributed but logically centralized" data management across 19 sectors and five innovation areas, including public security, urban governance, and social credit.

This aims to enhance governance systems and social control. The National Dataset Management Service Platform, trialed April 29, 2026, aggregates government data for AI model training, with state-supervised data exchanges tracking flows. "High-quality datasets" are technically defined (SAC/TC609, August 2025) to be large, secure, and politically "correct," explicitly excluding content violating socialist core values. By June 2025, over 35,000 such datasets, exceeding 400 petabytes, were built. This initiative integrates economic development with the Party's social and political control objectives, embedding control in data quality and sharing mechanisms.

No comments: