China proposes blacklist of training data for generative AI models - ET Telecom
China has introduced strict security requirements for generative AI firms, including a blacklist prohibiting training data containing illegal or harmful content. This regulation mandates that models cannot utilize information deemed detrimental to national stability or social harmony, effectively enforcing existing internet censorship standards on artificial intelligence development. The proposal emphasizes ethical compliance by requiring explicit consent for using personal and biometric data, while also outlining guidelines to prevent intellectual property violations. These measures aim to balance innovation with user rights, ensuring that AI systems respect individual privacy and legal ownership as they evolve within the domestic market. This initiative is crucial to open data discourse as it illustrates how state-level governance directly restricts data accessibility for technological advancement. By legally defining permissible training sources, these regulations highlight the tension between open data principles and national security imperatives, potentially fragmenting global AI development and influencing how international datasets are utilized in regulated jurisdictions.
Source: telecom.economictimes.indiatimes.comPublished on 2023-10-14