Nvidia, Magic AI and Chinese companies MiniMax and 01.ai are joining major US artificial intelligence developers such as Google, OpenAI and Microsoft by signing up to commitments for responsible AI development and deployment agreed at a summit in Seoul last May. The additions bring a significant increase in the number of Chinese companies represented, just ahead of a follow-up summit in Paris next week.
Nvidia, Magic AI and Chinese companies MiniMax and 01.ai are joining major US artificial intelligence developers such as Google, OpenAI and Microsoft by signing up to commitments for responsible AI development and deployment agreed at a summit in Seoul last May.A document update today by the UK government, which cohosted the South Korean summit, has revealed a significant increase in Chinese companies on the multinational list, which already features France's Mistral AI, South Korea’s Naver, Abu Dhabi’s G42, sees (see here).
The voluntary "Frontier AI Safety Commitments" include one on publishing safety frameworks, including how developers will assess risks of their general AI models and in what circumstances they may not deploy a model.
The commitments also require developers to define when severe risks are “deemed intolerable” and, in extreme circumstances, to “not develop or deploy a model or system at all” if mitigations cannot keep risks below certain thresholds. The expectation was that the signatories would share their efforts ahead of the Paris AI Action Summit, which begins on Monday.
Alibaba and Tencent-backed MiniMax develops multimodal large language models and character-based chatbots. 01.ai specializes in open-source large language models and was founded by Taiwanese investor Kai-fu Lee, previously an executive at both Google and Microsoft.
Previously the only Chinese company on the original list of 16 was Zhipu.ai, another large language model developer with backing from Alibaba companies, Tencent and Xiaomi.
Magic AI, based in California, is explicitly developing artificial general intelligence, or AGI, via research and code automation.
Nvidia, principally a hardware developer and dominant in the design of chips used for training and running AI, joins Samsung Electronics on the now-growing list of members.
Nvidia has had a turbulent few weeks, with its share price plummeting after publicization of a generative AI app by Chinese company DeepSeek. The sophisticated, free-to-use model is reported not to have needed the most powerful chips, as supplied by Nvidia.
The OECD today launched a new reporting tool with the buy-in of some of the Frontier AI Safety Commitments signatories, including OpenAI, Microsoft, Amazon, Anthropic and Google. The tool allows developers to upload their efforts around safety and transparency as per the G7’s Hiroshima AI Code of Conduct (see here).
The UK technology ministry did not respond to requests for comment.
Please e-mail editors@mlex.com to contact the editorial staff regarding this story, or to submit the names of lawyers and advisers.