China publica una lista negra de datos para entrenar modelos de IA generativa
China has introduced stringent security requirements for generative AI companies, notably banning content deemed illegal or harmful from training datasets. This regulatory framework mandates rigorous safety evaluations for public-facing models and enforces strict data protection standards, including obtaining consent for personal data usage and safeguarding intellectual property rights. These measures reflect a shift towards tighter control following earlier, more lenient regulations, aiming to prevent the circulation of information that undermines social stability or state interests. By defining specific thresholds for prohibited content, the government ensures that AI development aligns closely with national political and social objectives while managing the rapid commercial rollout of these technologies. This development is highly relevant to open data because it demonstrates how national security concerns can restrict data accessibility and reuse. It highlights the tension between open innovation and regulatory compliance, showing that even widely available datasets may be subject to exclusion based on political criteria. Such restrictions challenge the global open data ethos by prioritizing state control over unrestricted information flow in critical technological sectors.
Source: finanzasdigital.comPublished on 2023-10-13
Related news
- China proposes blacklist of training data for generative AI models
- RTI Day celebrated - OrissaPOST
- Taiwanese AI taken down after it repeats Chinese government line
- Millions of Workers Are Training AI Models for Pennies
- Transparency group unveils proposal to enshrine Arkansas FOIA in state Constitution