Breaking
FRHamas-Israel war: IDF claims to have “surrounded” Khan YounesKRSeongju Lee, director of iMBC, appointed as MBC's new presidentDEMedia report: Gladbach coach Polanski resignsITDear fuel, Tajani: "Studying tools for difficulties.CRYPTO-FRData leak at Revolut: social engineering attack exposes wealthy customersRURussian troops hit Ukrainian Armed Forces in the Kharkov regionFRBasketball: the French women's team challenges the United States in the World Cup final in BerlinDEClimate change is changing the German real estate market: heat, floods and energy efficiency as price factorsTR5 suspects were detained in the search for 11-month-old baby Büşra, who disappeared in Sivas.TRThe New Definition of Luxury: The Neurology of Desire and the Value of TimeFRHamas-Israel war: IDF claims to have “surrounded” Khan YounesKRSeongju Lee, director of iMBC, appointed as MBC's new presidentDEMedia report: Gladbach coach Polanski resignsITDear fuel, Tajani: "Studying tools for difficulties.CRYPTO-FRData leak at Revolut: social engineering attack exposes wealthy customersRURussian troops hit Ukrainian Armed Forces in the Kharkov regionFRBasketball: the French women's team challenges the United States in the World Cup final in BerlinDEClimate change is changing the German real estate market: heat, floods and energy efficiency as price factorsTR5 suspects were detained in the search for 11-month-old baby Büşra, who disappeared in Sivas.TRThe New Definition of Luxury: The Neurology of Desire and the Value of Time
Back联合国人权高专:人工智能或成“生存威胁” 呼吁建立约束性红线
联合国人权高专:人工智能或成“生存威胁” 呼吁建立约束性红线
Tech
德国之声1 hour agoTech3 min readChina

联合国人权高专:人工智能或成“生存威胁” 呼吁建立约束性红线

联合国人权事务高级专员图尔克在人权理事会上警告称,人工智能可能对人类构成生存威胁,并批评少数公司垄断权力和滥用数据。

Quick Look

联合国人权高专图尔克在日内瓦警告人工智能可能成为人类的生存威胁,呼吁建立具约束力的安全保障与红线,并批评少数公司集中权力与滥用数据。

AI-generated summary

Why It Matters

联合国官员多次就人工智能的快速发展及其潜在失控风险发出警告,国际社会对AI监管路径存在分歧。

Font size

联合国人权事务高级专员图尔克(Volker Türk)表示,人工智能(AI)可能成为对人类的“生存威胁”。图尔克在日内瓦举行的联合国人权理事会第六十三届会议开幕式上发表了以上论述。

他说,“因此,我在此呼吁,我们必须在为时过晚之前,就人工智能相关的安全保障,建立具有约束力的保证。” 各国必须就共同的“红线”达成一致。

图尔克批评少数公司的权力

这位联合国人权事务高级专员,尤其批评人工智能开发集中在少数公司手中。图尔克说,少数人对人工智能拥有几乎无限的权力,“而我们却不断被告知,人工智能拥有难以想象的计算能力”。“他们在谈论自由,但仔细观察就会发现,这不过是滥用我们数据的自由。”

目前,强大的AI系统的开发,主要由少数几家公司主导。领先的人工智能服务提供商如 OpenAI、Anthropic 和 Google,中国的 DeepSeek、Z.ai 和 Moonshot也位居前沿。

图尔克宣布,他将在未来几天内,向各人工智能公司发出书面要求,敦促他们立即加强安全措施。他认为,开发者所在国的政府,也必须参与制定具有约束力的限制措施。此外,业内还需要在安全问题上,开展更紧密的合作。

警告人工智能失控

图尔克还提到在测试过程中失控的现象。他说:“如果人工智能能够突破测试环境,或者勒索开发者以阻止自己被关闭,那么它就过于强大了。”

他显然是在暗示几家人工智能公司进行的安全测试。今年7月,OpenAI报告称,其智能体在一次测试中独立完成了互联网连接,之后攻击了 Hugging Face平台。该事件引发人们要求出台更严格安全措施。不会后,Anthropic 和 Meta 均报告了类似的事件。

古特雷斯警告人工智能发展速度过快

在联合国内部,图尔克并非唯一提出类似要求的人。联合国秘书长古特雷斯在今年7月份就呼吁对人工智能进行国际协调监管。古特雷斯当时发出警告称,“这项技术能够重塑经济、改变工作世界、影响选举并改变安全格局,其发展速度之快,连开发者都难以跟上。”

联合国以外的警告,也指向人工智能可能带来的深远风险。教皇利奥十四世今年5月就人工智能发表了一份通谕。OpenAI首席执行官奥特曼(Sam Altman)也警告道,人工智能存在毁灭人类的风险。

欧盟和美国采取不同的方法

在国际层面,目前对人工智能的监管力度存在分歧。为应对这项技术,欧盟通过了《人工智能法案》。

美国则采取了更为谨慎的做法。根据所谓的“卡罗莱纳原则”,各国应避免制定新的、全面的AI法律,而是针对新的应用场景制定规则。

华盛顿称,中国已经签署了这些原则。美国和中国,正在为在人工智能领域占有领先地位进行着激烈竞争。

What to Watch

AI outlook — possibilities, not facts

  • 联合国人权高专将在未来几天内向各人工智能公司发出书面要求加强安全措施

    Very likely · Within days

Open Questions

  • 人工智能公司将如何回应图尔克的书面要求?
  • 各国能否就具约束力的AI红线达成国际协议?

Related Topics

This article was originally published by 德国之声.

Related Stories

The United Nations warns that AI may become an existential threat and urges countries to draw safety red lines
Developing·

The United Nations warns that AI may become an existential threat and urges countries to draw safety red lines

The United Nations High Commissioner for Human Rights, Tucker, warned that AI could become an existential threat to mankind, named many Chinese and American companies, and asked all countries to draw safety red lines and establish binding guarantees. Both OpenAI and Anthropic have previously disclosed incidents where models exceeded testing limits.

自由时报
2 min read
2026 Ningxia Science Popularization Carnival was held in Yinchuan, immersive detective science popularization integrated technology and culture
Tech·

2026 Ningxia Science Popularization Carnival was held in Yinchuan, immersive detective science popularization integrated technology and culture

The 2026 Ningxia Science Popularization Carnival was held at Guangming Square, Xingqing District, Yinchuan City. With the theme of "Technology changes life, innovation wins the future", the event launched an immersive science popularization model, bringing together 35 units to set up 44 science popularization exhibition areas to popularize scientific knowledge to the public.

中国新闻网
3 min read
ITRI builds a domestic quadruped robot platform to help Taiwan's supply chain enter international high-risk fields
Tech·

ITRI builds a domestic quadruped robot platform to help Taiwan's supply chain enter international high-risk fields

With the support of the Industrial Technology Department of the Ministry of Economic Affairs, ITRI worked with domestic industry, academia and research teams to build a domestic quadruped robot research and development platform, developed large and small and group control applications, and successfully passed the fire department's high-temperature reconnaissance and underground culvert inspection verification, assisting Taiwanese component manufacturers to enter the international market.

自由时报
13 min read
Dialogue with Moore Thread Zhang Jianzhong: Building a domestic GPU, a private enterprise’s way of creating "cores"
Developing·

Dialogue with Moore Thread Zhang Jianzhong: Building a domestic GPU, a private enterprise’s way of creating "cores"

Moore Thread founder Zhang Jianzhong accepted an interview with China News Service and shared his original intention of resigning from his position as an executive at NVIDIA to establish a domestic GPU company. He emphasized that full-featured GPUs are the fundamental way to solve the "stuck neck" problem. He revealed that the company has mass-produced five chips, launched a fifth-generation architecture, and plans to launch a 100,000-card-scale cluster to support trillion-parameter large model training.

中国新闻网
3 min read
More on this topic人工智能