DeepSeek invited to brief UN Security Council on AI risks — Reuters

Reuters reported, citing two sources, that DeepSeek will be among companies briefing the UN Security Council this week on AI risks, with Moonshot also invited, while OpenAI CEO Sam Altman and senior Anthropic representatives are also expected.

GOVERNMENT desk — a Chinese frontier lab sitting next to U.S. labs at the UN Security Council is the geopolitical AI story of the week.

The meeting is still ahead. Reuters says the 15-member Security Council is set to meet on Wednesday to discuss AI and international security. The Security Council is the UN body charged with international peace and security. Fifteen members is the council’s full voting membership. The wire is dated Tuesday, 22 Sep, and it says Wednesday. It does not print a calendar date for the session. “Is set to meet” and “will be among the companies briefing” are future tense. This filing does not say the session has happened, and it does not say a company has already spoken.

Who else Reuters places on the list. OpenAI chief executive Sam Altman plans to brief the meeting. Diplomats also expect senior representatives of Anthropic. A diplomat, here, is a government’s representative at the UN. Reuters says it reported that lineup last week. File Altman’s plan and the Anthropic expectation as Reuters. A plan to brief is not a transcript. An expectation is not a confirmation that those representatives have taken a seat.

The Chinese invitations, still one source via Reuters. DeepSeek and other Chinese AI firms, including Moonshot, have been invited to make statements. DeepSeek’s founder, Liang Wenfeng, does not plan to attend. The same source said plans remain fluid and could change at the last minute. Fluid means the list can still move. File the invitation, the founder’s plan not to attend, and that caveat as the source, via Reuters. Do not turn “does not plan to attend” into a claim that Liang Wenfeng is in the chamber, or that a final absence is already locked. The meeting has not happened.

DeepSeek and Moonshot did not immediately respond to Reuters’ requests for comment. No reply is not a confirmation, and it is not a denial this desk can quote. This filing does not invent a statement from either company.

Reuters also sets the week, and those lines stay scene. World leaders are in New York for the annual UN General Assembly. The wire says the session comes weeks after several AI industry leaders called for a coordinated slowdown, warning that systems could soon improve on their own and slip beyond human control. It notes the council first discussed AI risks in 2023, a warning last week from UN Secretary-General Antonio Guterres, and that U.S. Treasury Secretary Scott Bessent said on Monday that senior U.S. and Chinese officials agreed to hold further talks on AI safety. File that scene as Reuters’. It is not a claim that DeepSeek has already briefed the council.

Plain English for the rest of the card: Security Council = the 15-member UN body for international peace and security. exclusive = Reuters’ label that it has the story first. two sources = people familiar with the plan, unnamed on the page. briefing = a statement to the council. It has not happened yet. Moonshot = another Chinese AI firm the source says was invited. founder = Liang Wenfeng, who the source says does not plan to attend. fluid = the plan can still change. no comment = DeepSeek and Moonshot had not answered Reuters when the story was published. This filing is that report. It is not a UN document, and it is not a company statement.

REPORTED here: Reuters’ 22 Sep 2026 exclusive by Laurie Chen — Tier B wire, two unnamed sources, not a UN document and not a DeepSeek or Moonshot statement. The invitation for DeepSeek to brief the Security Council this week on AI risks, the two sources, the Wednesday session of the 15-member council on AI and international security, Sam Altman’s plan to brief, the diplomatic expectation of senior Anthropic representatives, the Moonshot invitation, Liang Wenfeng’s plan not to attend, the fluid-plans caveat, and the no immediate comment from DeepSeek and Moonshot are Reuters-attributed. NOT claimed: that the briefing already happened, that Liang Wenfeng attended or will attend, a locked absence, a DeepSeek or Moonshot comment, an Altman transcript from the council, a seat chart, a UN primary this desk holds, a stock tip, or investment advice. Distinct from the already-filed un-ai-panel-agents, openai-frontier-standards, and bessent-china-ai-dialogue.