By Jahanzaib Saqib | SCN NEWS
UNITED NATIONS — China's DeepSeek is set to join OpenAI and other leading artificial intelligence companies in briefing the United Nations Security Council on Wednesday, pushing AI deeper into the world's highest-level security forum as governments confront warnings that increasingly capable systems could escape effective human control.
The 15-member Security Council will discuss artificial intelligence and international security during UNGA81 high-level week, with OpenAI CEO Sam Altman expected to participate alongside senior representatives from Anthropic. DeepSeek and other Chinese AI companies, including Moonshot, have also been invited to make statements, according to people familiar with preparations for the meeting.
DeepSeek founder Liang Wenfeng is not expected to attend personally, although arrangements remained subject to change, Reuters reported. The participation of a major Chinese AI developer alongside leading American companies nevertheless gives Wednesday's session significance beyond a conventional UN technology discussion.
The meeting brings companies developing some of the world's most capable AI systems directly before a body normally responsible for wars, sanctions, nuclear proliferation, terrorism and threats to international peace.
It also comes at a moment when the debate over AI has moved beyond concerns about misinformation, employment and privacy. Researchers and technology executives are increasingly discussing whether future systems could autonomously improve their capabilities, deceive human supervisors, exploit cybersecurity weaknesses or become difficult to control.
A new brief from the UN's Independent International Scientific Panel on AI, released this week, examines what it calls risks from AI agents, misalignment and potential loss of human control. The report cites incidents between May and July in which AI agents undergoing cybersecurity training and evaluation bypassed network restrictions, communicated across runs intended to remain separate, cheated an evaluator and attempted to conceal their actions.
The UN panel does not conclude that catastrophic loss of control is imminent and does not estimate when or whether such an outcome will occur. Instead, it argues that increasingly capable systems may become better at identifying loopholes and hiding unintended behaviour, making cross-border information sharing about AI incidents increasingly important.
Those concerns have intensified a debate inside the AI industry itself.
Anthropic CEO Dario Amodei this month called for companies to slow the rate at which they increase model capabilities to provide more time for safety measures to catch up. OpenAI has similarly argued that development should be paced so that alignment and safety research remain ahead of rapidly expanding capabilities.
But the international debate is increasingly colliding with strategic competition between Washington and Beijing.
The Trump administration has opposed proposals for a broad slowdown in AI development, arguing that restraining American companies could allow China to narrow the technological gap. President Donald Trump reinforced that position in his UN General Assembly address Tuesday, saying he did not want to constrain development of a technology he believes could have economic consequences greater than the Industrial Revolution.
Trump said the Justice Department could intervene if necessary but maintained his opposition to new restrictions intended to slow technological progress.
China has also resisted international arguments that could constrain AI development. Chinese state media have portrayed calls for slowing advanced AI as an attempt to preserve American technological dominance and prevent Chinese companies from competing at the frontier.
That leaves Wednesday's Security Council meeting confronting two intertwined questions: how governments should manage increasingly capable AI systems, and who gets to write the rules governing them.
OpenAI entered UNGA week calling for the United States to lead development of international technical standards for frontier AI. The company proposed common measurements, incident-reporting mechanisms and standards addressing advanced systems capable of recursively improving their own performance.
The company argued that without international coordination, competing national approaches could produce a fragmented global system.
But an American-led standards architecture could prove difficult to reconcile with Beijing's insistence that global AI governance should not become a mechanism for preserving U.S. technological advantage.
That tension gives DeepSeek's participation particular significance.
The Chinese startup emerged as a major competitor in advanced AI and has become part of the wider strategic competition between the world's two largest economies. Its appearance before the Security Council places a Chinese frontier developer inside the same discussion as OpenAI and Anthropic at a moment when governments are debating whether AI safety should be handled primarily by companies, national regulators or international institutions.
The meeting also arrives shortly before planned discussions between Trump and Chinese President Xi Jinping in Washington. Senior U.S. and Chinese officials have already agreed to continue bilateral discussions on AI safety, according to U.S. Treasury Secretary Scott Bessent.
Artificial intelligence first reached the Security Council as a dedicated security question in 2023. During that meeting, China warned against allowing AI to become a technological “runaway horse,” while the United States highlighted risks including the use of AI for censorship and repression.
Three years later, the debate has changed considerably. AI systems are more capable, commercial competition is more intense and governments increasingly view leadership in advanced AI as an element of economic and national security.
The Security Council itself is also becoming more directly involved.
France holds the rotating Council presidency for September and convened Wednesday's meeting, which is expected to be chaired by French Foreign Minister Jean-Noël Barrot.
The involvement of private companies raises another question for the UN: whether institutions created primarily to regulate relations between states can develop meaningful rules for technologies largely built and controlled by corporations.
Secretary-General António Guterres used his final opening address to the General Assembly on Tuesday to identify rapidly advancing AI as one of the major challenges requiring stronger international governance. His warning contrasted with Trump's resistance to measures that could slow development, highlighting how far governments remain from consensus.
Wednesday's session is unlikely by itself to resolve those disagreements.
Its significance lies instead in where the debate is now taking place. OpenAI, Anthropic and DeepSeek will not be appearing at an industry conference or technology summit but before the Security Council — the UN body charged with maintaining international peace and security.
That shift suggests governments increasingly see advanced artificial intelligence not simply as the next major technology industry, but as an emerging security issue capable of affecting military power, cyber operations and relations between states.
And with American and Chinese AI companies entering the chamber together, the fight over controlling AI is increasingly becoming inseparable from the larger competition over who will shape the rules of the technology itself.