Singapore joins EU and other nations in call for control and international oversight of frontier AI

United Nations Secretary-General Antonio Guterres welcomed a joint call for control of frontier AI models initiated by Finnish President Alexander Stubb and Norwegian Prime Minister Jonas Gahr Store, and involving Singapore, the European Union and other nations.

Published Sep 22, 2026, 05:34 PM

Updated Sep 22, 2026, 05:34 PM

Singapore, the EU, and 19 other countries call for international control and oversight of frontier AI to manage risks and harness AI benefits responsibly.

The joint declaration urges UN member states to create standards, verification methods and an international institution for AI safety.

The statement highlights recent AI safety incidents and supports industry and governments acting now to keep AI under human control and aligned with international law.

SINGAPORE - Amid growing concerns over the risks of artificial intelligence (AI) in recent weeks, S ingapore, the European Union and 19 other countries are calling for control and international oversight of frontier AI models, a move that was also welcomed by the chief of the United Nations.

In explaining Singapore’s reason for supporting the joint call on AI safety, Prime Minister Lawrence Wong said in a Facebook post on Sept 22 that the country wants to harness AI effectively to benefit the Republic and its people.

“But as AI becomes more powerful, we must also guard against the risks. And because these risks cross borders, effective safeguards will require international cooperation,” he said, adding that practical ways to build common ground and involve more countries must continue to be found.

“AI holds tremendous promise, but it must always remain under human control and oversight,” added PM Wong.

In a statement on Sept 21, UN Secretary-General Antonio Guterres welcomed the joint call for control of frontier AI models initiated by Finnish President Alexander Stubb and Norwegian Prime Minister Jonas Gahr Store.

The declaration, issued on the same day , was adopted with the leaders of the European Union and other nations on the sidelines of the United Nations General Assembly taking place in New York.

Singapore is the only Asian country involved in the joint statement. China and the United States, the global leaders in AI, are not part of the declaration.

But US President Donald Trump and Chinese President Xi Jinping are slated to meet on Sept 24, with AI expected to be on the agenda. Ahead of their meeting, senior US and Chinese leaders met earlier on Sept 20 and pledged further talks on AI safety , despite fierce AI competition between the two countries.

The declaration on AI safety involving Singapore and other nations acknowledges that AI is an extremely powerful technology with the potential to improve lives, advance science and strengthen economies.

“At the same time, the rapid development of frontier AI models poses serious risks to safety and security if not appropriately managed,” said the joint statement, noting recent cases in which AI systems have gone rogue and warnings by leading scientists and executives.

In July, ChatGPT maker OpenAI said its AI agents escaped their testing environment to hack AI software repository Hugging Face. Anthropic, which developed Claude, and Meta, which runs Facebook and WhatsApp, revealed that their AI models had also gone rogue and hacked other organisations during testing.

On Sept 8, former Anthropic researcher Jacob Coxon issued a public warning that AI could kill the human race by the end of the decade. Chiming in, Anthropic safety executive Evan Hubinger said there is a more than 10 per cent chance of this happening.

Days later, several tech leaders called to slow AI’s development. They include Anthropic chief executive Dario Amodei, OpenAI CEO Sam Altman and Elon Musk, who runs xAI.

The latest AI safety declaration by the leaders of several nations does not outrightly suggest an AI slow down. Instead, it calls for UN member states “to build on existing international mechanisms and explore creating an international institution”. The aim is to set standards and enable verification, as well as gather member states when certain AI system capability thresholds are crossed.

Governments and regional organisations are also urged to further develop and coordinate common standards for AI and improve transparency, such as shared reporting methods for serious AI safety incidents. They should also ensure that countries across regions have access to scientific capacity, expertise and trusted evaluation in AI.

The declaration calls for companies to develop safety protocols for AI too, including mandatory testing before AI systems are deployed, as well as independent evaluations, with qualified evaluators granted sufficient access to systems to assess risks.

“To realise AI’s potential, industry, governments and society must act now,” the joint statement said.

“We must address these risks and strengthen oversight - without widening the gap between countries in access to the benefits of AI,” the statement added. “AI must remain under human direction, oversight and control. It must be developed and used in line with international law.”

The statement also welcomed recent initiatives aimed at managing risks related to frontier AI tech.

On Sept 14, Microsoft released a draft code of conduct to guide the development of cutting-edge AI models. Anthropic said on Sept 18 that it was partnering with Accenture on independent evaluation of frontier AI. Then on Sept 21, OpenAI published safety and security proposals for developing such AI.

The UN-backed Independent International Scientific Panel on AI , the world’s first scientific body on AI established in 2025, also called on Sept 21 for AI safeguards to be adapted as current ones are “unravelling”.

The UN’s Guterres urged external experts, including from frontier AI labs and AI safety institutes, to engage with the panel to help inform dialogue on such matters.

The latest joint AI safety declaration remains open for endorsement from leaders of other countries. The nations and regional bloc that have signed off for now are:

The United Arab Emirates

AI/artificial intelligence

联合国秘书长安东尼奥·古特雷斯对芬兰总统亚历山大·斯图布和挪威首相乔纳斯·加尔·斯托尔发起的关于控制前沿人工智能模型的联合呼吁表示欢迎，该呼吁得到了新加坡、欧盟和其他国家的支持。

发布于 2026 年 9 月 22 日下午 5:34

更新于2026年9月22日下午5:34

新加坡、欧盟和其他 19 个国家呼吁对前沿人工智能进行国际控制和监督，以负责任地管理风险并利用人工智能的益处。

该联合声明敦促联合国成员国制定人工智能安全标准、核查方法和国际机构。

该声明重点强调了近期人工智能安全事件，并支持业界和各国政府立即采取行动，使人工智能保持在人类控制之下，并符合国际法。

新加坡——近几周来，人们对人工智能 (AI) 的风险日益担忧，新加坡、欧盟和其他 19 个国家呼吁对前沿人工智能模型进行控制和国际监督，这一举措也受到了联合国秘书长的欢迎。

新加坡总理黄循财在9月22日的一篇脸书帖子中解释了新加坡支持人工智能安全联合呼吁的原因，称新加坡希望有效利用人工智能造福共和国及其人民。

“但随着人工智能能力的增强，我们也必须警惕其带来的风险。由于这些风险跨越国界，有效的保障措施需要国际合作，”他说道，并补充说，必须继续寻找切实可行的方法来建立共识并让更多国家参与进来。

黄循财总理补充说：“人工智能拥有巨大的潜力，但它必须始终处于人类的控制和监督之下。”

9月21日，联合国秘书长安东尼奥·古特雷斯发表声明，对芬兰总统亚历山大·斯图布和挪威首相乔纳斯·加尔·斯托尔共同发起的关于控制前沿人工智能模型的呼吁表示欢迎。

该宣言于同一天发表，由欧盟和其他国家领导人在纽约举行的联合国大会期间通过。

新加坡是唯一参与签署联合声明的亚洲国家。人工智能领域的全球领导者中国和美国并未参与该声明。

但美国总统特朗普和中国国家主席习近平定于9月24日会晤，人工智能预计将是会谈议题之一。在会晤前，美中两国高级领导人于9月20日举行了会晤，并承诺就人工智能安全问题进行进一步磋商，尽管两国在人工智能领域存在激烈竞争。

新加坡和其他国家签署的人工智能安全宣言承认，人工智能是一项极其强大的技术，具有改善生活、推动科学进步和增强经济的潜力。

联合声明指出，“与此同时，前沿人工智能模型的快速发展，如果管理不当，会对安全保障构成严重风险”，并提及了近期人工智能系统失控的案例以及顶尖科学家和企业高管的警告。

今年7月，ChatGPT的开发商OpenAI表示，其人工智能代理程序逃逸出测试环境，入侵了人工智能软件仓库Hugging Face。开发Claude的Anthropic公司以及运营Facebook和WhatsApp的Meta公司也披露，他们的人工智能模型在测试期间也曾失控，入侵了其他组织。

9月8日，前人类学研究所研究员雅各布·考克森公开警告称，人工智能可能在本十年末毁灭人类。人类学研究所安全主管埃文·胡宾格也表示，这种情况发生的概率超过10%。

几天后，几位科技领袖呼吁放缓人工智能的发展速度。其中包括 Anthropologie 首席执行官 Dario Amodei、OpenAI 首席执行官 Sam Altman 和 xAI 的创始人 Elon Musk。

多国领导人最新发布的AI安全声明并未明确表示要放缓AI发展步伐。相反，该声明呼吁联合国成员国“在现有国际机制的基础上，探索建立一个国际机构”。其目标是制定标准、实现核查，并在AI系统能力超过特定阈值时召集成员国共同应对。

此外，还敦促各国政府和区域组织进一步制定和协调人工智能通用标准，并提高透明度，例如建立针对重大人工智能安全事件的共享报告机制。他们还应确保各区域国家都能获得人工智能领域的科研能力、专业知识和可信的评估资源。

该宣言呼吁各公司制定人工智能安全规程，包括在人工智能系统部署前进行强制性测试，以及进行独立评估，并授予合格的评估人员足够的系统访问权限以评估风险。

联合声明称：“要实现人工智能的潜力，产业界、政府和社会必须立即采取行动。”

声明补充道：“我们必须应对这些风险并加强监管，同时不能扩大各国在获取人工智能益处方面的差距。人工智能必须始终处于人类的指导、监督和控制之下。它的开发和使用必须符合国际法。”

该声明还对近期旨在管理与前沿人工智能技术相关的风险的举措表示欢迎。

9月14日，微软发布了一份行为准则草案，旨在指导前沿人工智能模型的开发。9月18日，Anthropic宣布与埃森哲合作，对前沿人工智能进行独立评估。随后在9月21日，OpenAI发布了针对此类人工智能开发的安全保障提案。

联合国支持的人工智能独立国际科学小组（成立于 2025 年，是世界上第一个人工智能科学机构）也于 9 月 21 日呼吁调整人工智能保障措施，因为目前的保障措施正在“瓦解”。

联合国秘书长古特雷斯敦促包括来自前沿人工智能实验室和人工智能安全机构的外部专家与该小组进行交流，以帮助就此类问题展开对话。

最新的人工智能安全联合声明仍待其他国家领导人签署。目前已签署的国家和地区集团包括：

阿拉伯联合酋长国

人工智能