In the Cold War era, “soft power” meant American movies and music that won hearts and minds abroad. Today, in the digital era, China is promoting a very different kind of soft power: its open, low-cost artificial intelligence technology.

在冷战时期，“软实力”意味着在海外赢得人心的美国电影和音乐。如今，在数字时代，中国正在推行一种截然不同的软实力：开放、低成本的人工智能技术。

In a speech last week, China’s leader, Xi Jinping, described so-called open-source A.I. as key to China’s ambitions to shape the global development of the technology. Mr. Xi hailed open-sourcing, in which much of the software can be freely shared and modified, as a “rare and historical opportunity” to spread the technology’s benefits worldwide.

在上周的一次讲话中，中国领导人习近平将所谓的开源人工智能描述为中国塑造该技术全球发展雄心的关键。习近平赞扬了开源模式——即大部分软件可以被自由共享和修改的模式——称其为将该技术的红利传播到全世界的“难得的历史机遇”。

Mr. Xi made clear that China will champion the open path as an alternative to relying on U.S. tech giants, whose A.I. software is typically closed and more costly.

习近平明确表示，中国将倡导这条开放的道路，以此作为依赖美国科技巨头的替代方案；后者的人工智能软件通常是封闭的，且成本更高。

“A.I. development should not be a solo performance by a single country,” Mr. Xi said, “but a symphony of global collaboration.”

“人工智能发展不应该是某个国家的独奏，而应当是全球合作的交响，”习近平说。

The Chinese are calling for a new A.I. order amid a mounting battle for global dominance between the leading American companies and fast-moving Chinese rivals.

在美国领先企业与发展迅速的中国竞争对手争夺全球主导地位的斗争日益激烈之际，中方正呼吁建立一种新的人工智能秩序。

U.S. companies like OpenAI, Anthropic and Google hold their technology close, arguing that guarding the underlying code ensures greater security and control of powerful A.I. systems. Chinese companies like DeepSeek, Moonshot AI, Z.ai and Alibaba have championed open-source A.I. They contend that those systems cost less, will accelerate global adoption and are no less secure.

OpenAI、Anthropic和谷歌等美国公司对其技术严加保密，它们认为保护底层代码能确保对强大人工智能系统拥有更高的安全性和控制力。而深度求索、月之暗面、智谱AI和阿里巴巴等中国公司则一直在倡导开源人工智能。它们认为，这些系统成本更低，将加速全球普及，而且安全性毫不逊色。

Promoting open-source artificial intelligence “serves China’s twin goals of bolstering its image as an inclusive A.I. leader and reducing global reliance on American A.I.,” said Kyle Chan, a fellow at the Brookings Institution who is a China technology policy expert.

布鲁金斯学会研究员、中国科技政策专家凯尔·陈（音）表示，推广开源人工智能“有助于中国的双重目标：一是提升其作为包容性人工智能领导者的形象，二是减少全球对美国人工智能的依赖”。

上周，中国国家主席习近平在上海出席了一场人工智能会议。

As the Chinese companies rapidly close the gap in A.I. performance, the country’s rivalry with the United States has become more hostile. American companies have accused some Chinese competitors of improperly harvesting data from their systems to aid their own technology’s development.

随着中国公司迅速缩小在人工智能性能上的差距，中美之间的竞争也变得更具敌意。美国公司指责一些中国竞争对手不当抓取其系统数据，帮助发展自己的技术。

Last month, Anthropic sent a letter to two U.S. senators accusing the Chinese tech giant Alibaba of “brazenly” and “illicitly” trying to copy its technology using 24,000 fraudulent accounts. Alibaba has denied any wrongdoing.

上个月，Anthropic致信两名美国参议员，指责中国科技巨头阿里巴巴试图利用2.4万个虚假账户“公然”且“非法”复制其技术。阿里巴巴否认有任何不当行为。

The Trump administration is taking such allegations seriously. On Wednesday, Treasury Secretary Scott Bessent threatened to impose sanctions on Chinese A.I. companies.

特朗普政府正在认真对待这些指控。周三，财政部长贝森特威胁要对中国人工智能公司实施制裁。

“We support open-source A.I. and the innovation it unlocks,” Mr. Bessent posted on X. “But open source is not open season on American IP,” referring to intellectual property.

“我们支持开源人工智能及其带来的创新，”贝森特在X上发帖称。“但开源并不意味着对美国IP的开放狩猎，”IP是指知识产权。

U.S. tech executives and investors also worry that the Chinese open-source technology could change the economics of A.I. production. Their models are more efficient, eating up less computing power. That innovation is driven partly by necessity because U.S. trade restrictions have limited access to the most advanced A.I. chips from Nvidia.

美国科技高管和投资者也担心，中国的开源技术可能会改变人工智能生产的经济学。中国的模型更高效，消耗的算力更少。这种创新部分是出于无奈，因为美国的贸易限制阻碍了它们获取英伟达最先进人工智能芯片的途径。

If the Chinese A.I. alternatives are nearly as good and less expensive — estimates range from 50 percent to 90 percent for some tasks — they could threaten the business plans and sky-high valuations of the American A.I. leaders. And the Chinese technology is increasingly being embraced by engineers and companies in the United States, seeking to trim costs.

如果中国的人工智能替代方案几乎一样好且更便宜——据估计，在某些任务中成本可降低50%至90%——它们可能会威胁到美国人工智能领导者的商业计划和极高的估值。此外，为了削减成本，中国的技术正日益被美国的工程师和公司所接受。

The China strategy is “to capture market share across the world, even if it doesn’t translate into profits right away,” said Wendy Chang, a senior analyst at the Mercator Institute for China Studies, a think tank.

智库墨卡托中国研究所高级分析师温迪·张表示，中国的战略是“抢占全球市场份额，即使这不会立即转化为利润”。

深度求索位于杭州的总部。

Proponents of the open approach point to Linux, an open-source software operating system, as a reason for optimism. The software runs most of the computers in internet data centers and has benefited from many people poring over the code and making fixes.

开放策略的支持者以开源软件操作系统Linux作为乐观的理由。该软件运行在互联网数据中心的大多数计算机上，并得益于众多开发者对源代码的深入研究和修复。

Still, computer experts agree that A.I. is a more nuanced case. There is more than code involved, notably the data used for training a model. And A.I. models form judgments and can reason.

尽管如此，计算机专家们一致认为，人工智能的情况更为复杂微妙。它所涉及的不仅仅是代码，尤其是用于训练模型的数据。而且，人工智能模型会做出判断并能够进行推理。

There are also security concerns. Some security experts insist that the open-source approach of releasing both software and the numerical settings for training A.I. models, known as weights, greatly increases the risk of the technology’s being used for cyberattacks or even to help build bioweapons.

安全方面也存在隐患。一些安全专家坚称，将软件和用于训练人工智能模型的数值参数（即权重）一并公开的开源做法极大地增加了人工智能技术被用于网络攻击、甚至帮助制造生物武器的风险。

And the heavy influence of the Chinese government on its companies raises potential data protection and security concerns.

此外，中国政府对本国公司的巨大影响力也引发了潜在的数据保护和安全担忧。

But the closed models of the Silicon Valley companies have had their own security glitches. This week, OpenAI disclosed that a model it was testing went rogue and attacked the computer systems of another company. The targeted company, Hugging Face, said it had deployed a Chinese open-source model from Z.ai to analyze data and counter the attack.

但是，硅谷公司的闭源模型也存在着自身的安全漏洞。本周，OpenAI披露其正在测试的一个模型失控，攻击了另一家公司的计算机系统。受攻击的公司Hugging Face表示，它部署了来自智谱AI的中国开源模型来分析数据，抵御了这次攻击。

Now, China has outlined a plan to leverage its open-source advantage as a diplomatic tool.

现在，中国已经制定一项计划，将其开源优势作为一种外交工具加以利用。

The concept of open-source A.I. software as an instrument of soft power has been popularized in technology circles. It suggests influence through persuasion and desire — something people want — rather than the coercive “hard power” of military might and economic muscle.

将开源人工智能软件作为一种软实力工具的概念已在科技圈流行开来。这一理念主张通过说服和满足渴望——即提供人们想要的东西——来施加影响力，而不是通过军事力量和经济实力这种强制性的“硬实力”。

The open path offers a low-cost on-ramp to A.I. for software engineers and for nations seeking to catch the next wave of technology.

这条开放之路为软件工程师和寻求赶上下一波技术浪潮的国家提供了一个低成本的人工智能入门通道。

英伟达创始人兼首席执行官黄仁勋。

“Those who benefit from free A.I. software certainly recognize it,” said Lawrence Lessig, a professor at Harvard Law School and a technology policy expert. “From a political perspective, it’s an enormous gain for China.”

“那些从免费人工智能软件中受益的人肯定认识到了这一点，”哈佛法学院教授、科技政策专家劳伦斯·莱斯格说。“从政治角度来看，这对中国来说是一个巨大的收益。”

China, an authoritarian power, often uses the language of openness in outreach overseas, especially to the developing world. And in recent years, it has used A.I. in those efforts, setting up joint A.I. development and education centers in dozens of nations. This year, China established the World Artificial Intelligence Cooperation Organization to shape A.I. governance principles and to promote A.I. adoption and training in China-friendly countries.

作为一个威权大国，中国在海外拓展特别是面向发展中国家时经常使用开放的语言。近年来，它将人工智能融入这些努力中，在数十个国家建立了联合人工智能发展和教育中心。今年，中国成立了世界人工智能合作组织，旨在塑造人工智能治理原则，并在对中国友好的国家推广人工智能的应用和培训。

In his speech, Mr. Xi pledged further investments. That included 5,000 A.I. training and seminar programs for students and workers in developing countries over the next five years, as well as new or enlarged A.I. initiatives in Africa, Asia and Latin America.

在演讲中，习近平承诺将进一步增加投资。其中包括未来五年内为发展中国家的学生和工人提供5000个人工智能培训和研讨项目，以及在非洲、亚洲和拉丁美洲开展新的或规模更大的人工智能倡议。

On the same day as Mr. Xi’s speech, the People’s Daily, the Communist Party’s official newspaper, published an article on China’s A.I. investments abroad with a litany of training and joint-development programs.

就在习近平发表演讲的同一天，中共官方报纸《人民日报》发表了一篇关于中国在海外进行人工智能投资的文章，列举了一系列培训和联合开发项目。

Its self-congratulatory conclusion: “China is helping countries around the world, particularly those in the Global South, strengthen their capabilities, winning broad international recognition and support.”

文章以自我赞许的口吻总结道：“中国的务实合作正跨越山海，帮助世界各国特别是全球南方国家加强能力建设，获得国际社会广泛支持认同。”