MIKE ISAAC2026年9月14日Anthropic首席执行官达里奥·阿莫迪，摄于2月。他在一篇文章中写道：“过去几个月里，我逐渐确信，要完全应对风险需要更加审慎。” Ludovic Marin/Agence France-Presse — Getty ImagesDario Amodei, the chief executive of Anthropic, called for a global slowdown of artificial intelligence development in a 3,800-word essay on Saturday, just days after one of the company’s employees quit over concerns about the safety of the technology.Anthropic首席执行官达里奥·阿莫迪周六发表了一篇3800词的文章，呼吁全球放缓人工智能发展。就在几天前，该公司的一名员工因担忧人工智能技术的安全性辞职。Top executives of other major A.I. companies quickly signed on to the effort. Sam Altman, the chief executive of OpenAI; Elon Musk, whose SpaceX rocket company has been ramping up spending on A.I.; and Demis Hassabis, the chair of Google DeepMind, wrote in social media posts that they agreed on the need for a slower pace.其他主要人工智能公司的高管迅速响应这一呼吁。OpenAI首席执行官萨姆·奥尔特曼、Space X（这家火箭公司一直在加大对人工智能的投入）的埃隆·马斯克，以及谷歌DeepMind董事长戴密斯·哈萨比斯在社交媒体上发帖，表示他们同意需要放慢步伐。Mr. Amodei, who co-founded Anthropic to focus on securely and carefully building A.I., wrote in his essay that while he believed the technology could bring many benefits, it was advancing at too rapid a pace for researchers to continue safely.阿莫迪是Anthropic的共同创始人，致力于安全、谨慎地构建人工智能。他在文章中写道，尽管他相信这项技术可以带来许多好处，但其推进速度过快，研究人员无法继续安全地发展。“Over the last few months, I have become convinced that fully addressing the risks requires even more prudence — not just investing in risk prevention, but pacing the rate of capabilities advancement so that risk prevention has time to keep up,” Mr. Amodei wrote. “We must slow the pace at which we improve the capabilities of A.I. models. Progress will still seem fast, and we must make wise use of the time we gain.”“过去几个月里，我逐渐确信，要完全应对风险需要更加审慎——不仅是投资于风险预防，还要调节能力迭代的速度，以便风险预防有时间跟上，”阿莫迪写道。“我们必须放慢改进人工智能模型能力的速度。进展仍将看起来很快，我们必须明智地利用争取到的时间。”Mr. Amodei has warned about the safety risks of A.I. for years. But his essay amounted to one of the most full-throated calls for urgent action from a leading A.I. company executive to date.阿莫迪多年来一直警告人工智能的安全风险。但在所有领先人工智能公司的高管中，他的文章构成了迄今为止最全力呼吁采取紧急行动的声音之一。He called for independent auditors to check the companies for safety standards and urged global cooperation in creating rules around A.I. development.他呼吁独立审计师检查公司的安全标准，并敦促全球合作制定围绕人工智能发展的规则。The appeal for an industrywide pause follows months of escalating concerns across the technology industry as A.I. model development has made significant strides at companies like Anthropic, the developer of the Claude chatbot, as well as OpenAI, the creator of ChatGPT, and Google’s DeepMind.他发出这一全行业暂停的呼吁之际，Anthropic（Claude聊天机器人的开发者）、OpenAI（ChatGPT的创造者）和谷歌DeepMind等公司在人工智能模型开发方面取得了重大进展，数月来科技行业的担忧也不断升级。Tensions rose further after the leading A.I. labs discovered security breaches carried out by their own programs without their knowledge. In one incident this year, OpenAI’s systems breached an A.I. start-up called HuggingFace, which was not detected until HuggingFace informed OpenAI weeks after it had happened. Since then, A.I. researchers have conferred to discuss what to do.在领先的人工智能实验室发现其自身程序在它们不知情的情况下利用安全漏洞后，紧张局势进一步加剧。在今年的一起事件中，OpenAI的系统入侵了一家名为HuggingFace的人工智能初创公司，直到HuggingFace在事件发生数周后通知OpenAI才被发现。此后，人工智能研究人员召开会议讨论应对之策。On Tuesday, Jacob Coxon, an Anthropic researcher, resigned and posted a series of social media posts about how the leading A.I. labs were not responsibly developing the technology and were gambling with people’s lives, setting off a furor. Other A.I. researchers have called on their leaders to pause A.I. development, while some are forming splinter groups to raise more awareness about the risks of A.I.周二，Anthropic研究员雅各布·科克森辞职并发布了一系列社交媒体帖子，讲述领先的人工智能实验室如何没有负责任地开发这项技术，并在拿人们的生命赌博，引发轩然大波。其他人工智能研究人员呼吁他们的领导人暂停人工智能发展，而一些人正在组建分裂团体以提高对人工智能风险的认识。A.I. leaders are girding themselves for a day when an A.I. system can rely on itself to grow more advanced rather than requiring the aid of human researchers. This dynamic, which computer scientists call “recursive self improvement,” is what some top A.I. developers believe may lead to the software spinning out of their control.人工智能领导人正在为某一天做好准备，届时人工智能系统可以自主进化，不需要人类研究人员的帮助。计算机科学家将这种动态称为“递归自我改进”，一些顶级人工智能开发者认为这可能导致软件失控。“Left unchecked, it could outrun our ability to understand and control these systems, and so must be pursued very carefully, if at all,” Mr. Amodei said.“如果任其发展，它可能超出我们理解和控制这些系统的能力，因此必须非常谨慎地推进，如果真要推进的话，”阿莫迪说。Hours after Mr. Amodei published his essay, Mr. Altman posted a statement to social media agreeing that a slowdown of A.I. development was necessary.在阿莫迪发表文章数小时后，奥尔特曼在社交媒体上发布声明，同意放缓人工智能发展是必要的。“This has been a primary topic of discussions we’ve had at OpenAI in recent weeks,” Mr. Altman wrote. “Committing to having independent evaluators with employee-like access is a great idea, and we will do the same. We’ll have more to share soon.”“这是OpenAI最近几周讨论的主要话题，”奥尔特曼写道。“承诺让独立评估者拥有类似员工的访问权限是一个好主意，我们也会这样做。我们很快会有更多分享。”Mr. Musk, in a post on his social media platform X, said simply, “Dario is right.”马斯克在其社交媒体平台X上简明地发帖表示，“达里奥是对的。”On Saturday evening, Mr. Hassabis said Mr. Amodei’s essay was a step in the right direction. “The details need working through, but the direction is correct for meeting this critical moment,” Mr. Hassabis posted on social media.周六晚间，哈萨比斯表示阿莫迪的文章是朝着正确方向迈出的一步。“细节需要梳理，但方向对于迎接这一关键时刻是正确的，”哈萨比斯在社交媒体上发帖写道。For years, many of the top A.I. labs have been locked in bitter competition with one another. But as these systems have advanced, there has been some broader agreement among executives at rival companies that something needs be done.多年来，许多顶级人工智能实验室一直陷于彼此之间的激烈竞争。但随着这些系统的进步，竞争对手公司的高管之间已达成一些更广泛的共识，即需要采取一些行动。Last month, as OpenAI prepared to release an advanced A.I. model called Astra, the company said it had intentionally slowed down its pace of development, citing advancements in cybersecurity capabilities. In July, Google’s Mr. Hassabis published an open framework for what he believed would be a safe way to cooperatively develop next-generation A.I.上个月，准备发布一款名为Astra的先进人工智能模型的OpenAI表示已有意放慢发展步伐，理由是网络安全能力的进步。7月，谷歌的哈萨比斯发布了一个开放框架，阐述他认为合作开发下一代人工智能的安全方式。(The New York Times has sued OpenAI and Microsoft, claiming copyright infringement of news content related to A.I. systems. The two companies have denied those claims.)（《纽约时报》已起诉OpenAI和微软，主张其与人工智能系统相关的新闻内容存在版权侵权。两家公司已否认这些指控。）There are few rules on A.I. development, though President Trump signed an executive order in June asking for A.I. labs to voluntarily submit their models for review before public release.人工智能发展几乎没有规则约束，尽管特朗普总统在6月签署了一项行政命令，要求人工智能实验室在公开发布前自愿提交其模型接受审查。In his essay on Saturday, Mr. Amodei suggested actions that the industry might take to slow the pace of development.在周六的文章中，阿莫迪提出了行业可能采取的行动，以放慢发展步伐。Mr. Amodei said that all A.I. labs could agree to third-party technology assessments from “embedded evaluators,” or outside specialists who could verify best safety practices across companies. He also suggested that countries with democratic governance systems coordinate to create safety standards, which could take the form of regulatory action.阿莫迪说，所有人工智能实验室都可以同意接受来自“嵌入式评估者”的第三方技术评估，即可以验证各公司最佳安全实践的外部专家。他还建议拥有民主治理体系的国家协调制定安全标准，这可以采取监管行动的形式。He added that it would probably require a global effort working with other nations, including authoritarian ones, to properly coordinate a slowdown.他补充说，这可能需要与包括威权国家在内的其他国家合作，进行全球努力，才能适当协调放缓。Not everyone in Silicon Valley believes that Mr. Amodei’s intentions are purely altruistic. Some critics have said that Anthropic and OpenAI are looking to cement their position as A.I. market leaders by stoking fear and asking the federal government to create regulatory frameworks that would make it more difficult for smaller companies to compete with them.并非硅谷所有人都认为阿莫迪的意图是纯粹利他的。一些批评者说，Anthropic和OpenAI正寻求通过煽动恐惧，并要求联邦政府建立监管框架，以巩固其人工智能市场领导者的地位，这些框架将使较小的公司更难与它们竞争。At a tech conference in San Francisco this past week, Jensen Huang, the chief executive of the A.I. chip maker Nvidia, said that A.I. labs were ratcheting up security concerns because they planned to introduce new cybersecurity products.在上周于旧金山举行的一次科技会议上，人工智能芯片制造商英伟达的首席执行官黄仁勋说，人工智能实验室正在加剧安全担忧，因为它们计划推出新的网络安全产品。“What better way to create demand than to create a problem?” Mr. Huang said at the event. “Who doesn’t want their market to be hysterical about their product and line up around the corner for it?”“还有什么比制造问题更好的创造需求的方式？”黄仁勋在活动中说。“谁不希望自己的产品让自己的市场陷入狂热，并排队购买？”On Saturday, shortly after Mr. Amodei published his essay, Clément DeLangue, the chief executive of HuggingFace, said in a social media post that it was imperative that A.I. safety issues not “be solved behind the closed doors of a handful of frontier labs.” Nvidia agreed to buy HuggingFace this month for $12.9 billion.周六，在阿莫迪发表文章后不久，HuggingFace首席执行官克莱门特·德朗格在社交媒体帖子中说，人工智能安全问题绝不能“由几家前沿实验室关起门来自行解决”。英伟达本月同意以129亿美元收购HuggingFace。In his essay, Mr. Amodei stressed that he still finds A.I. capable of bringing “incredible benefits” to humanity, including potentially curing diseases and accelerating economic growth. But even so, he said the risks of A.I. were too great to not proceed with extreme caution.在他的文章中，阿莫迪强调，他仍然认为人工智能能够为人类带来“不可思议的好处”，包括可能治愈疾病和加速经济增长。但即便如此，他说人工智能的风险太大，务必要极其谨慎地推进。“The measures I propose to advance the frontier at a safe pace will not be easy,” Mr. Amodei wrote. “But I believe we owe it to humanity to try.”“我提出的以安全步调推进前沿的措施执行起来不会容易，”阿莫迪写道。“但我相信，为了人类，我们应该尝试。”Mike Isaac是《纽约时报》硅谷记者，常驻旧金山。他报道全球最具影响力的科技公司，以及它们如何塑造互联网和现实世界的文化。翻译：经雷点击查看本文英文版。