The National Information Security Standardization Technical Committee (TC260) is seeking public input on a proposed national standard titled "Foundational Security Criteria for Generative Artificial Intelligence Services in Cybersecurity Technology."

This draft standard lays out essential security requirements for generative artificial intelligence (AI) services, covering areas like data security during training, model integrity, and protective measures. It's aimed at adoption by service providers involved in security assessments and could become a reference for regulatory bodies.

Moreover, the draft standard comes with an appendix detailing the main security risks related to training data and AI-generated outputs.

To access the press release, click here and download the draft national standard, click here. Note that both resources are only available in Chinese.