KATE CONGER2026年9月10日Anthropic旧金山办公室外的抗议活动。 Lucas Foglia for The New York TimesFor years, the artificial intelligence industry raced to improve its technology as quickly as possible. Now some researchers are increasingly sounding the alarm about the need to slow down.多年来，人工智能行业竞相以最快速度改进其技术。如今，一些研究人员正日益就放慢速度的必要性发出警告。Jacob Coxon, an Anthropic researcher who previously worked at OpenAI, said on Tuesday that he had resigned from Anthropic because “neither company is acting responsibly.” The two leading A.I. labs are sprinting to build “superhuman systems that can hack anything, revolutionize any field overnight, and acquire real power and resources” without building in appropriate safeguards, he said in a social media post.曾任职于OpenAI的Anthropic研究人员雅各布·考克森周二表示，他已从Anthropic辞职，因为“两家公司都没有负责任地行事”。他在一则社交媒体帖子中说道，这两家领先的人工智能实验室正在急切地构建“能够侵入任何地方、一夜之间彻底改变任何领域、并获取真正权力和资源的超人系统”，却没有内置适当的保障措施。Mr. Coxon’s warning echoed concerns that other A.I. experts have voiced in recent months. Some A.I. leaders have long believed that the technology could create doomsday scenarios if left unchecked, and their concerns heightened after OpenAI’s models broke out of its systems in July and hacked Hugging Face, an A.I. model library.考克森的警告呼应了其他人工智能专家近几个月表达的担忧。一些人工智能领导人长期以来一直认为，如果放任不管，这项技术可能制造末日情景。在OpenAI的模型于7月突破其系统并入侵Hugging Face（一个人工智能模型库）之后，他们的担忧加剧了。That same month, more than 1,300 employees at top A.I. companies, including Anthropic, OpenAI, Meta and Google’s DeepMind, signed an open letter calling on the U.S. government to create rules to slow A.I. development. In August, more than 100 major tech companies committed to providing their best A.I. models to organizations like hospitals and infrastructure providers so they could prepare for A.I.-powered cyberattacks.同月，包括Anthropic、OpenAI、Meta和谷歌DeepMind在内的顶级人工智能公司的1300多名员工签署了一封公开信，呼吁美国政府制定规则以放慢人工智能发展。8月，超过100家大型科技公司承诺向医院和基础设施提供商等组织提供其最好的人工智能模型，以便它们能够为人工智能驱动的网络攻击做好准备。OpenAI also said last month that it would expand safety testing and slow the release of a new model, Astra, which has powerful cybersecurity abilities.OpenAI上个月还表示，将扩大安全测试，并延缓新模型Astra的发布，该模型具有强大的网络安全能力。In an interview on Wednesday, Mr. Coxon said incidents like the Hugging Face hack had heightened his concerns that A.I. models were advancing too quickly. When he looked a year or two ahead, he was “viscerally worried about what things will look like,” Mr. Coxon said. “It was mostly a gradual emotional thing and eventually it kind of snapped, and then I decided to leave.”考克森周三接受采访时表示，像Hugging Face被入侵这样的事件加剧了他的担忧，即人工智能模型推进得太快。当他展望一两年后，他“发自内心地担心事情会变成什么样”，考克森说。“这主要是一个逐渐累积的情绪过程，最终有点像是崩了，于是我决定离开。”Some critics have argued that A.I. companies are overhyping the technology to make their products appear more valuable. Several lawmakers said on Wednesday, however, that they were taking Mr. Coxon’s warning seriously.一些批评者认为，人工智能公司夸大了这项技术，以使其产品显得更有价值。然而，几位议员周三表示，他们正在认真对待考克森的警告。“There are massive implications of a race towards super intelligence,” Representative Anna Paulina Luna of Florida, a Republican, wrote in a social media post in which she called on Congress to convene a special session focused on A.I. regulation. “Not enough of Congress is focusing on this.”“急切奔向超级智能会带来深远的后果，”佛罗里达州共和党众议员安娜·保利娜·卢纳在一则社交媒体帖子中写道，她在帖子中呼吁国会召开一次专注于人工智能监管的特别会议。“国会对此关注得不够。”“Safety researchers are resigning, powerful A.I. models are breaking out of their labs, and companies are racing ahead anyway,” Representative Lori Trahan of Massachusetts, a Democrat, wrote in a social media post. “It’s past time for Congress to get off the sidelines and do its job.”“安全研究员正在辞职，强大的人工智能模型正在突破其实验室，而企业仍然不管不顾地竞相前进，”马萨诸塞州民主党众议员洛里·特拉汉在社交媒体发帖写道。“国会早就该停止袖手旁观，开始做好自己的分内事了。”While leading A.I. labs have agreed that the government should place restrictions on the pace at which the technology is developed and released to the public, many companies are fiercely competing with one another and fear falling behind their rivals if they pause.尽管领先的人工智能实验室一致认为，政府应该对这项技术开发和向公众发布的速度施加限制，但许多公司正在激烈竞逐，并担心一旦停下来就会落后于对手。“They are locked in a race to get there first — they believe no one else will act responsibly, so they must do it themselves, despite the risk,” Mr. Coxon said in his social media post.“他们陷入了一场争当第一的竞赛——他们相信没有谁会负责任地行事，所以他们必须自己来做，尽管有风险，”考克森在他的社交媒体帖子中说道。At least one Anthropic employee agreed with Mr. Coxon’s assessment.至少有一名Anthropic员工同意考克森的评估。“We really do earnestly believe A.I. could kill all humans!” Evan Hubinger, who works at Anthropic on creating A.I. that aligns with human ethics and values, wrote on social media. Mr. Hubinger said he believed the risk of A.I.’s eliminating humanity was greater than 10 percent within the next decade.“我们确实真心相信人工智能可能杀死全人类！”埃文·胡宾格在社交媒体上写道，他在Anthropic工作，致力于创建与人类伦理和价值观对齐的人工智能。胡宾格表示，他相信人工智能消灭人类的风险在未来十年内超过10%。“I believe Anthropic is trying its best, but we do not yet have a plan to solve alignment for superintelligence and are not clearly on track to,” he added.“我相信Anthropic正在尽力而为，但我们还没有解决超级智能对齐问题的计划，也没有找到明确的解决路径，”他补充道。An Anthropic spokeswoman said on Wednesday that the company had long been transparent about the benefits and risks of A.I., and was building models with some of the strongest safeguards in the industry.一位Anthropic发言人周三表示，该公司长期以来一直对人工智能的好处和风险保持透明，并且正在构建拥有业内最强保障措施的模型。“We believe the world would benefit from the industry adopting a lawful, verifiable way to work together to pace how we release powerful models,” the spokeswoman said.“我们相信，如果整个行业采用一种合法、可验证的方式共同协作，来调节我们发布强大模型的速度，世界将从中受益，”这位发言人说。OpenAI did not immediately respond to a request for comment about Mr. Coxon’s comments. Mr. Hubinger also did not immediately respond to a request for comment.OpenAI没有立即回应就考克森言论置评的请求。胡宾格也没有立即回应置评请求。(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。两家公司否认这些指控。）Dylan Freedman对本文有报道贡献。Kate Conger是一名科技记者，常驻旧金山。可以通过电子邮件kate.conger@nytimes.com联系她。翻译：经雷点击查看本文英文版。