OpenAI, Anthropic CEOs call for global AI regulation at UN

Industry leaders urge global regulation, as the Trump administration baulks at setting up new guardrails.

OpenAI CEO Sam Altman speaks during a UN Security Council meeting on artificial intelligence during the 81st United Nations General Assembly at the UN headquarters in New York, on September 23, 2026 [Angela Weiss/AFP]

The heads of several major AI firms told the United Nations Security Council (UNSC) their industry urgently needed global oversight to avoid dangers that could threaten the whole world.

“If managed poorly, I even believe AI could be a risk to humanity as a whole,” Dario Amodei, the chief executive officer of Anthropic, told members of the body on Wednesday.

list 1 of 3 As AI leaders warn of catastrophe, US and China shun slowdown calls

list 2 of 3 As Xi meets Trump, who’s winning their trade war?

list 3 of 3 US lawmakers propose sweeping AI restrictions with superintelligence ban

Sam Altman, the head of rival company OpenAI, echoed his concerns, telling the 15-member council tasked with tackling major crises globally that humanity could “lose control of the future of AI”.

The meeting, which coincides with the UN General Assembly (UNGA) gathering in New York City, was convened by France and comes at a time when experts are increasingly warning that the rapid development of AI needs more human oversight to ensure it does not slip out of control and cause a global catastrophe.

Altman and Amodei called on world leaders to take action.

“If AI is to be democratic, the most important decisions cannot be made by labs in San Francisco alone,” Altman told members. “They must be shaped through democratic processes and by governments accountable to the people they serve.”

Their concerns were shared by several representatives on the council, including the foreign ministers of France and the United Kingdom, who said the international community needed to step in and create common frameworks for how the technology should be controlled.

Hugging Face CEO Clement Delangue, whose company has come under attack by out-of-control AI models in recent months – incidents used by the other companies as evidence of the need for more safety measures – told the UNSC his company had relied on the technology to defend itself in those same incidents.

Delangue said Hugging Face had relied on a Chinese AI model to help defend against the attack by OpenAI’s AI agents, because it faced fewer restrictions than comparable US tools.

“We were attacked by AI, but more importantly, we defended ourselves with AI,” he told the council.

US and China reluctant to impose restrictions

In the United States, though, where the largest and most influential companies developing AI are based, the administration of US President Donald Trump has baulked at imposing new guardrails on the industry.

The administration’s representative at the UNSC meeting, Michael Kratsios, told members, “We totally reject all efforts by international bodies to assert centralised control and global governance of AI.”

Chinese President Xi Jinping is expected to discuss whether and how to regulate AI during a visit to Washington, DC, this week. The two countries are locked in a technological race to develop more powerful AI tools, a competition that experts say makes it less likely that either country would want to impose any major new restrictions on their efforts right away.

Yet there is a growing recognition at the UN of the danger AI potentially poses to the world, said Daniel Forti, head of UN Affairs at the International Crisis Group. Member states understand that “there will be much more of a need for international cooperation, setting some rules of AI, even if the biggest players are more focused on growth opportunities than on some sort of collaboration,” Forti said.

For several years, the UN has been participating in multilateral meetings to shape everything from protections for workers from AI in emerging economies and ensuring open access to this technology, to following how AI is used in military conflicts. In 2024, the UNGA unanimously passed its first resolution on AI, a nonbinding statement that called on member states to protect personal data, monitor AI for risks and safeguard human rights.

The adoption of AI has taken off dramatically since then, and with it have come dire warnings from environmental groups, human rights advocates, and even the tech moguls whose companies are developing the tech.

The future of AI “cannot be decided by a handful of countries or left to the whims of a few billionaires”, UN Secretary-General Antonio Guterres said at a global summit held earlier this year.

Last year, the UNGA formed two new bodies to deal with AI: the Independent International Scientific Panel on AI that brings together experts to provide governments with independent assessments, and the Global Dialogue on AI Governance, which provides a regular forum for discussing approaches to AI governance.

“The dangers are real and imminent,” Yoshua Bengio, a Canadian expert on AI and co-chair of the Independent International Scientific Panel, told the UNSC on Wednesday. “This council faces an unprecedented threat, one that none of its members would ⁠choose, that none can contain alone, and that does not respect the borders we defend.”

行业领袖们呼吁制定全球监管措施，而特朗普政府却不愿设立新的监管机制。

OpenAI 首席执行官 Sam Altman 于 2026 年 9 月 23 日在纽约联合国总部举行的第 81 届联合国大会期间，在联合国安理会关于人工智能的会议上发表讲话。[Angela Weiss/AFP]

几家主要人工智能公司的负责人向联合国安理会表示，他们的行业迫切需要全球监管，以避免可能威胁全世界的危险。

“如果管理不善，我甚至认为人工智能可能会对整个人类构成威胁，”人类组织首席执行官达里奥·阿莫迪周三告诉该组织成员。

列表 1/3：人工智能领袖警告灾难之际，美国和中国却拒绝放缓发展步伐

列表 2/3：习近平会见特朗普，谁将赢得这场贸易战？

美国三位议员提议对人工智能进行全面限制，包括禁止超级智能，名单如下。

竞争对手 OpenAI 的负责人 Sam Altman 也表达了同样的担忧，他告诉负责应对全球重大危机的 15 名成员组成的委员会，人类可能会“失去对人工智能未来的控制”。

此次会议与在纽约举行的联合国大会同期举行，由法国召集。此时，专家们越来越警告说，人工智能的快速发展需要更多的人类监督，以确保它不会失控并造成全球灾难。

奥特曼和阿莫迪呼吁世界各国领导人采取行动。

“如果人工智能要走向民主，最重要的决策就不能仅仅由旧金山的实验室做出，”奥特曼告诉成员们。“这些决策必须通过民主程序制定，并由对人民负责的政府来制定。”

他们的担忧也得到了安理会几位代表的认同，其中包括法国和英国的外交部长，他们表示国际社会需要介入，并为如何控制这项技术建立共同框架。

Hugging Face 首席执行官 Clement Delangue 向联合国安理会表示，他的公司近几个月来一直受到失控人工智能模型的攻击——其他公司利用这些事件作为需要更多安全措施的证据。

Delangue 表示，Hugging Face 依靠中国人工智能模型来帮助防御 OpenAI 人工智能代理的攻击，因为与类似的美国工具相比，中国人工智能模型受到的限制较少。

“我们遭到了人工智能的攻击，但更重要的是，我们利用人工智能进行了自卫，”他告诉委员会。

美国和中国都不愿实施限制

然而，在美国——人工智能领域规模最大、最具影响力的公司所在地——美国总统唐纳德·特朗普的政府一直不愿对该行业施加新的限制。

美国政府在联合国安理会会议上的代表迈克尔·克拉齐奥斯告诉成员们：“我们完全拒绝国际机构为实现人工智能的集中控制和全球治理所做的一切努力。”

中国国家主席习近平预计将在本周访问华盛顿期间讨论人工智能监管问题及其监管方式。中美两国正竞相开发更强大的人工智能工具，专家表示，这种竞争使得两国都不太可能立即对彼此的人工智能研发成果施加任何重大限制。

国际危机组织联合国事务负责人丹尼尔·福蒂表示，联合国越来越认识到人工智能可能对世界构成的危险。福蒂说，成员国都明白，“即使最大的人工智能企业更关注自身发展机遇而非合作，未来也更需要国际合作，制定一些人工智能规则。”

多年来，联合国一直参与多边会议，致力于制定各项政策，包括保护新兴经济体劳动者免受人工智能侵害、确保开放获取这项技术，以及监督人工智能在军事冲突中的应用。2024年，联合国大会一致通过了其首个关于人工智能的决议，这是一项不具约束力的声明，呼吁成员国保护个人数据、监测人工智能风险并维护人权。

从那时起，人工智能的应用迅速发展，随之而来的是来自环保组织、人权倡导者，甚至是那些开发这项技术的科技巨头的严厉警告。

联合国秘书长安东尼奥·古特雷斯在今年早些时候举行的一次全球峰会上表示，人工智能的未来“不能由少数几个国家决定，也不能任由少数亿万富翁一时兴起”。

去年，联合国大会成立了两个新的机构来处理人工智能问题：人工智能独立国际科学小组，该小组汇集专家，为各国政府提供独立评估；以及人工智能治理全球对话，该对话提供了一个定期论坛，讨论人工智能治理的方法。

“危险真实存在且迫在眉睫，”加拿大人工智能专家、独立国际科学小组联合主席约书亚·本吉奥周三在联合国安理会表示。“安理会面临前所未有的威胁，这是任何一个成员国都不愿看到的，任何一个成员国都无法独自应对，而且这种威胁并不尊重我们所捍卫的边界。”