DUSTIN VOLZ2026年9月22日上周，北京千寻智能公司总部数据训练中心里的员工。人工智能将成为本周联合国大会数场会议的主导议题。 Tingshu Wang/ReutersAs the threats posed by artificial intelligence spark international alarm, the stakes of the Chinese leader Xi Jinping’s state visit to Washington this week have only grown bigger.随着人工智能带来的威胁引发国际恐慌，中国领导人习近平本周对华盛顿的国事访问显得非同小可。Earning considerably less attention are the meetings focused on A.I. at the United Nations General Assembly in New York this week, as world leaders gather in the shadow of several rogue A.I. hacking episodes that have prompted greater fears.相比之下，本周在纽约联合国大会上聚焦人工智能的会议受到的关注要少得多。此时，各国领导人正聚集在引发更大担忧的几起失控人工智能黑客事件的阴影下。A.I. will be a dominant theme for a fleet of U.N. meetings on Monday centered on international digital cooperation. The schedule includes discussions about the technology’s risks and opportunities, the potential for economic dislocation and possible international governance.人工智能将成为周一围绕国际数字合作展开的一系列联合国会议的主导议题。议程包括讨论该技术的风险与机遇、潜在的经济错位以及可能的国际治理方案。Mr. Xi is expected to skip the General Assembly to instead convene in Washington with President Trump, who has repeatedly dismissed A.I. safety concerns as a hoax.预计习近平将缺席联合国大会，转而在华盛顿与特朗普总统会面，特朗普已多次将对人工智能安全的担忧斥为骗局。There is no shortage of tension between the two countries, and A.I. is just one of several issues the two sides are expected to discuss. But few issues are as contentious right now — or as important to humanity, some say — as the race for A.I. dominance.两国之间紧张对峙频频，且人工智能只是预计双方将讨论的众多问题之一。但一些人表示，眼下很少有哪项议题能像争夺人工智能主导权这样具争议性，或对人类如此重要。The United States has repeatedly accused Chinese A.I. firms of stealing the intellectual property of American A.I. labs. China, too, continues to agitate against U.S. restrictions on advanced computer chips needed to power A.I. models.美国已多次指责中国人工智能公司窃取美国人工智能实验室的知识产权。中国也持续抗议美国限制驱动人工智能模型所需的高级计算机芯片。Some onlookers, though expecting little, compared the urgency for a breakthrough on A.I. safety issues between the world’s two superpowers to the need for nuclear arms control agreements during the Cold War.尽管期望不高，但一些观察人士仍将两个超级大国在人工智能安全问题上取得突破的紧迫性与冷战期间缔结核军备控制协议的必要性相提并论。The U.N. gathering and the White House meeting underscore how much A.I. has become a race solely between two rival powers, with the rest of the world largely pushed to the periphery, foreign policy experts said.外交政策专家表示，联合国大会和白宫会晤凸显出，人工智能在很大程度上已演变为仅在两个竞争大国之间展开的角逐，而世界其他地区基本被推向边缘。“The U.N., to put it bluntly, has been sidelined in the A.I. conversation,” said Kyle Chan, a foreign policy fellow at the Brookings Institution who studies U.S.- China relations and focuses on technology issues.布鲁金斯学会研究美中关系并专注于科技问题的外交政策研究员陈凯欣(Kyle Chan)表示：“坦白说，联合国在人工智能对话中已被边缘化。”A.I. is “a two-player game,” Mr. Chan added. “It’s kind of like nuclear weapons: If you do not have them or are not on the verge of having them, it’s hard to be relevant in shaping the discussion.”人工智能是“两个玩家的游戏，”他还说。“这有点像核武器：如果你没有，或者不是即将拥有它们，就很难在塑造相关讨论时发挥实质性作用。”Reflecting that reality, the United States and China discussed on Sunday creating a notification system to allow for the countries to alert each other about A.I.-driven national security risks. The mechanism, Treasury Secretary Scott Bessent said, would be called “the U.S.-China A.I. dialogue.”同样反映这一现实的是，美中两国周日讨论了建立一个通报系统，以便两国互相通报由人工智能引发的国家安全风险。财政部长斯科特·贝森特表示，该机制将被命名为“美中人工智能对话”。It is not for lack of trying. António Guterres, the U.N. secretary general, pushed last week for international cooperation on A.I., highlighting it as one of three existential risks facing humanity alongside climate change and deepening global inequality.出现这种局面并非因为联合国未作尝试。联合国秘书长安东尼奥·古特雷斯上周大力推动人工智能领域的国际合作，强调它是与气候变化和不断加剧的全球不平等并列的三大人类生存风险之一。A.I., he said, presents other dangers, including amplifying misinformation, discrimination, privacy violations, cyberattacks and so-called superintelligence that could advance beyond human control.他指出，人工智能带来了其他危险，包括放大虚假信息、加剧歧视、侵犯隐私、引发网络攻击以及可能发展出脱离人类控制的所谓超级智能。“We need a shared understanding of how to advance the safe, secure and responsible development of A.I., while identifying when increasingly powerful systems may require stronger safeguards,” Mr. Guterres said. “The world cannot afford a race to the bottom on A.I. safety.”“我们需要就如何安全、可靠且负责任地推进人工智能发展达成共识，同时明确越来越强大的系统何时需要更强有力的保障措施，”古特雷斯表示。“世界承受不起在人工智能安全上竞相探底的后果。”OpenAI的Stargate是美国最大的数据中心项目，位于得克萨斯州阿比林。一位公司发言人表示，该公司首席执行官萨姆·奥特曼将于周三在联合国安全理事会发表演讲。Under Mr. Guterres, the United Nations has been trying to build international consensus around responsible A.I. development for years, including by forming an advisory body in 2023 to study the issue. And on Monday, the leaders of 22 countries adopted a declaration calling to strengthen “human direction, oversight and control” of A.I. systems. The United States and China did not join the declaration.在古特雷斯的领导下，联合国多年来一直致力于围绕负责任的人工智能发展凝聚国际共识，包括在2023年成立一个咨询机构专门研究该问题。周一，22个国家的领导人通过了一项宣言，呼吁加强对人工智能系统的“人类指导、监督和控制”。美国和中国均未加入该宣言。The two countries have only further cemented their leads ahead of everyone else since the U.N. began worrying about A.I., rushing to build out data centers and release cutting-edge A.I. models despite growing fears about existential risks associated with the technology.自联合国开始关注人工智能以来，美中两国相对其他国家的优势进一步扩大，竞相建设数据中心并发布尖端人工智能模型，尽管外界对该技术相关的生存风险日益担忧。In his remarks last week, Mr. Guterres acknowledged that “geopolitical divides” made policing A.I. difficult but said that voluntary pauses on the technology would not work if not everyone agreed to them — meaning, right now, the United States and China.在上周的讲话中，古特雷斯承认“地缘政治分歧”增加了监管人工智能的难度，但他表示，如果不能得到所有人的同意（这意味着目前必须包括美国和中国），暂停研发该技术的自愿呼吁将无法奏效。China had generally been seeking to increase its influence with the United Nations, but this summer launched the World Artificial Intelligence Cooperation Organization, which has its headquarters in Shanghai, in a bid to assert global dominance on international A.I. discussions. The United States has shown little interest in building international consensus on A.I. or other geopolitical matters.中国总体上一直在寻求扩大其在联合国的影响力，但在今年夏天推出了总部设在上海的世界人工智能合作组织，试图主导全球人工智能国际讨论。美国则对在人工智能或其他地缘政治问题上建立国际共识没有表现出多少兴趣。The U.N. Security Council may hold more sway on A.I. than the General Assembly. The OpenAI chief executive Sam Altman will speak to the Security Council on Wednesday, a company spokeswoman said. But it is difficult to imagine the Security Council — on which the United States, China and Russia sit — forging a consensus, experts said.联合国安全理事会在人工智能议题上可能比联合国大会更具影响力。一位公司发言人表示，OpenAI首席执行官萨姆·奥特曼将于周三在安理会发表演讲。但专家指出，很难想象有美、中、俄三国坐镇的安理会能够达成共识。Mr. Altman’s appearance at the United Nations is just a warm-up. A day later in Washington, he will join a far larger roster of A.I. and technology leaders for the U.S.-China state dinner, including Nvidia’s Jensen Huang, SpaceX’s Elon Musk and the Amazon founder Jeff Bezos.奥特曼在联合国的露面只是一场热身。一天后，他将在华盛顿与阵容更为庞大的人工智能和科技界领袖一同出席美中两国国宴，其中包括英伟达的黄仁勋、SpaceX的埃隆·马斯克以及亚马逊创始人杰夫·贝索斯。Julian Ku, a professor of constitutional and international law at Hofstra University, said that the United Nations has always struggled to influence superpowers on the uses and limits of weapons they have that others do not, from nukes to high-powered hacking tools.霍夫斯特拉大学宪法和国际法教授古举伦(Julian Ku)表示，从核弹到高性能黑客工具，在影响超级大国如何使用和限制它们独有的武器方面，联合国历来举步维艰。In the context of nuclear weapons, that took the form of helping nonnuclear nations gain access to nuclear power while spreading nonproliferation agreements, Mr. Ku said. While the United Nations would like to steer the A.I. safety conversation, it will probably find more success working to ensure the rest of the world is not fully left behind on the benefits societies could reap from the technology, Mr. Ku added.古举伦指出，在核武器背景下，联合国的努力体现为帮助无核国家获得核能，同时推广不扩散协议。他补充说，尽管联合国希望主导人工智能安全对话，但它可能会发现，不如转而致力于确保世界其他地区不会完全错过从该技术中获得的益处，这样效果会更好。“In a lot of ways, this is the story of the U.N. from the outset,” Mr. Ku said. “The U.N. can’t really shape the actions of those who have the power directly, but they do have the interests of everyone else who do not have that power.”“在许多方面，这就是联合国自诞生以来的写照，”古举伦说。“联合国无法直接左右那些权力者的行动，但它确实代表了不掌握这种权力的其他各方的利益。”Dustin Volz为时报撰写网络安全和情报方面的文章，他常驻华盛顿。翻译：纽约时报中文网点击查看本文英文版。