US-based AI companies like OpenAI, Anthropic have been secretly engaging with Chinese experts over AI Safety

In a rare display of collaboration between Chinese and American tech companies, leading artificial intelligence (AI) companies OpenAI, Anthropic, and Cohere have reportedly participated in clandestine diplomatic talks with Chinese AI experts, as per a report by the Financial Times. The meetings, which occurred in Geneva in July and October of 2023 aimed to address shared concerns about the potential spread of misinformation and threats to social cohesion posed by powerful AI technology. Insiders with direct knowledge revealed that scientists and policy experts from North American AI groups engaged in discussions with representatives from Tsinghua University and other Chinese state-backed institutions during the two Geneva meetings. The focus of the talks was to explore the risks associated with emerging AI technology and foster investments in AI safety research. The ultimate objective was to establish a scientific pathway for the safe development of more advanced AI technology. Participants emphasized the necessity for international standards on AI safety and alignment, highlighting the importance of agreement among key actors to facilitate broader consensus. These previously undisclosed talks mark a notable instance of Sino-US cooperation in the realm of cutting-edge technologies, especially amid the ongoing competition for supremacy in AI and quantum computing. The Geneva meetings, facilitated by the private mediation organization Shaikh Group, received support from the White House, as well as UK and Chinese government officials, according to an anonymous negotiator present at the discussions. The Chinese embassy in the UK expressed support for efforts to discuss AI governance and cooperation on global AI governance frameworks, norms, and standards. The talks, which included technical cooperation and policy proposals, contributed to discussions at the UN Security Council meeting on AI in July 2023 and the UK’s AI summit in November of the same year. Plans for future discussions are underway, focusing on scientific and technical proposals to align AI systems with legal codes, societal norms, and values. Notably, major Chinese AI companies such as ByteDance, Tencent, and Baidu did not participate, and Google DeepMind, while briefed on the talks, did not attend. Calls for cooperation between leading powers in addressing the challenges and risks associated with AI have been on the rise, with Chinese scientists and Western academics signing a statement in November advocating for tighter controls and international regulations on advanced AI systems. OpenAI, Anthropic, Cohere, and Tsinghua University declined to comment on their participation in the discussions. (With inputs from agencies)