In an unexpected move toward global AI governance, China and the US have partnered to develop the world’s first international standard for securing large language models in supply chains.
The development is the result of joint efforts between Chinese giants Ant, Baidu and Tencent, together with American companies Google, Meta and Microsoft.
The ‘Large Language Model Security Requirements for Supply Chain’ initiative was unveiled today at the World Digital Technology Academy (WDTA) in Shanghai.
Unexpected US, China AI Effort Revealed
The new standard aims to address the full lifecycle of LLMs to prevent security risks such as data leakage, model manipulation, and vendor non-compliance. Top academic and industry institutions, such as the Cloud Security Alliance Greater China Region and Nanyang Technological University in Singapore, joined the U.S. and Chinese companies in drafting and reviewing the guidelines. Together, the bodies form the AI Safety, Trust, and Responsibility (AI STR).
Peter Major, Chair of the United Nations Commission on Science and Technology for Development and Honorary Chair of the WDTA, noted: “International cooperation on AI-related standards has become increasingly important as artificial intelligence continues to develop and impact various sectors globally.”
In a blog postThe WDTA added: “This international collaboration is essential to managing the risks associated with AI while maximizing its benefits for all societies.”
As generative AI technologies continue to evolve, companies have called for greater measures to improve safety. OpenAI CEO Sam Altman previously called for “full-stack safety efforts.”
Despite US intentions to hamper Chinese technological and military advances with export restrictions, the two countries have cooperated in this area, underscoring the importance of establishing mutually acceptable and clear guidelines.
In addition, China became the first country in the world to have mainstream generative AI. Other countries and regions continue to lag behind.