United Nations Security Council Chamber at United Nations Headquarters in New York City Credit: Mikaela Ryden via Shutterstock According to an exclusive Reuters report published today, which cites two individuals familiar with the situation, DeepSeek is expected to speak to the UN Security Council this week about the risks posed by AI, alongside several US companies. The 15-member council will be meeting on Wednesday, while world leaders are in New York for the UN General Assembly. DeepSeek and another Chinese AI company, Moonshot, had not responded to Reuters’ requests for comment, and one of their sources warned that the plans could still be changed at the last minute. The same source says that invitations to speak have been extended to DeepSeek and several other Chinese AI companies, including Moonshot.
The source also notes that Liang Wenfeng, the founder of DeepSeek, is not expected to go himself. The Chinese laboratories will be joined by their American counterparts; Sam Altman, CEO of OpenAI, will give a briefing to the council in person, an OpenAI spokesperson told Reuters last week. Moreover, diplomats have informed the agency that they expect senior officials from Anthropic to take part as well. France, which has the rotating presidency of the council this month, is calling the meeting.
The session came after several weeks of public debate concerning how quickly frontier AI should advance. Dario Amodei from Anthropic had this month asked companies to slow the pace at which they improve their models, a suggestion that Altman and Elon Musk supported but Mark Zuckerberg from Meta opposed. As Reuters reported, governments are also divided: UN Secretary-General António Guterres told world leaders last week that the dangers posed by rapidly developing AI cannot be ignored, while US President Donald Trump maintained that existing safeguards are sufficient. The council began its discussion of AI in 2023, when China stated that AI must not become a ‘runaway horse’ and the United States expressed its concerns about governments using it for censorship and repression.
Also, a concept note that France has sent out for this week’s session highlights the potential for malicious use of AI and the danger of losing control over advanced models. In January 2025, DeepSeek, which is based in Hangzhou, attracted worldwide attention because the low cost of its R1 reasoning model had a damaging effect on US technology stocks. This week’s briefing will see one of China’s most well-known AI developers take part in the same UN discussion as the US research laboratories. Also tagged with
