China wants AI oversight to reach the data centre

Beijing’s proposed AI framework reaches from data and computing resources to models, deployment and applications. Companies would face stronger risk-management and incident-disclosure expectations, with regulatory sandboxes planned for sector-specific testing. The scope matters to foreign operators too: controlling compute and data can shape access to infrastructure before a product reaches a user. No draft law, thresholds or penalties have been published, so compliance teams have direction but not yet a rulebook.