2026年9月16日 / 美国东部时间下午12:04 / 哥伦比亚广播公司（CBS）新闻
随着人们愈发担忧，人工智能霸权竞赛可能将所有人推向一场堪比《终结者》电影末日情节的悲惨结局，分析师对CBS新闻表示，美中之间的相互猜忌，与这项技术本身一样，可能将人类“推下悬崖”。
一周前，Anthropic公司一名研究员做出的可怕预测登上了各大媒体头条，而包括他在内的专家们长期以来的担忧是，超级人工智能——一种理论上比任何人都更聪明的AI智能体——可能会将人类视为关键资源的竞争对手，从而危及我们的生存。
美国主要人工智能企业OpenAI和Anthropic的高管如今呼吁放缓人工智能开发速度，以“为前沿发展踩下刹车”。
作为与硅谷共同推动这一前沿领域发展的另一大枢纽，中国也发出了自身的警告。上周，一组美国研究人员利用AI曝光了中国“万能应用”微信的一处漏洞，这款应用在中国拥有超过8亿用户。
卡内基和平基金会非常驻学者王蕾娅表示，许多中国企业都认同越来越多的共识，即人工智能必须受到监管，她以上海人工智能实验室为例，称其是致力于人工智能安全的“极具声望的机构”。
北京人工智能安全与治理研究院院长易征在发给CBS新闻的一篇研究论文中写道：“人工智能带来的灾难性风险，以及人类灭绝的潜在危险，与日俱增。”
英国国王查尔斯三世将于周四在苏格兰召集行业领袖会议，讨论“能否制定一套共同原则”，以指导人工智能未来的开发与应用。
英伟达首席执行官黄仁勋的公司是此次会议的参会方之一，他在四月份曾表示，美中研究人员“切实开展对话”至关重要。
他说，行业领袖间的这种双边协调“如今严重缺失，因为我们当前将中国视为对手”。
猜忌与“中国杀人机器人”
一群美国民主党议员呼吁众议院议长迈克·约翰逊推迟即将到来的休会期，以便通过人工智能安全保障相关立法。
不过，约翰逊周日表示，仓促出台监管法规将意味着美国“会在竞赛中输给中国，这对每个美国人都是威胁”。
共和党参议员特德·克鲁兹上周表示：“如果真要出现杀人机器人，我宁愿它们是美国制造的，而非中国的。”
https://www.cbsnews.com/video/highlights-from-chinas-world-humanoid-games-and-what-it-means-for-the-future/
周一，特朗普总统将人工智能威胁的警告斥为“病态的阴谋论”，并在其Truth Social账号的帖子中补充道：“唯一对此感到高兴的是中国。谁赢得人工智能，谁就赢了！”
“不是美国政府不认为人工智能存在风险，”王蕾娅对CBS新闻说，而是在美国领导人“针对中国的人工智能思维中，首要关切是技术竞争”。
中国领导层已经发出了一些人工智能安全合作的信号。
今年6月，习近平主席明确警告了人工智能“失控”的危险。一个月后，他表示各国必须携手合作，“确保人工智能始终处于人类可控范围之内”。
但王蕾娅说，与许多美国高级官员一样，中国政府拒绝了放缓开发速度的呼吁，认为这一概念是美国“谋求技术霸权”的企图。
Anthropic首席执行官达里奥·阿莫代伊曾表示，美国及其全球伙伴应携手合作，“保持民主国家在人工智能领域对专制国家的尽可能大的领先优势，为我们争取有效踩刹车所需的喘息空间”。对此，中国外交部周一回应称，“威胁叙事……只会破坏全球人工智能治理进程”。
世界两大经济体之间的竞争与相互猜忌或许并不令人意外，但王蕾娅对CBS新闻表示，考虑到相关风险，将“人工智能安全对话简化为技术竞争对话”是“最无益的框架”。
新加坡国立大学贸易与技术专家亚历克斯·卡普里在2月份一篇题为《谁将拯救世界免于美中AI军备竞赛？》的文章中写道，两国“陷入了囚徒困境……如果双方都保持克制，都会更安全，但各自都在加速开发，以避免成为对方的受害者”。
前进之路
“协调放缓步伐或达成宏大的国际协议，或许并非吸引中国企业和政府参与的合适框架，”位于北京的AI安全咨询公司Concordia AI的国际治理主管关怡（Kwan Yee Ng）对CBS新闻表示。
不过她补充道，“两国可以在很多方面开展合作”。
据关怡所说，这些“唾手可得的成果”包括分享网络安全最佳实践、协调行动以避免恐怖组织或试图制造经济动荡的势力滥用人工智能，以及开启低级别的政府对话以共享细致的技术信息。
王蕾娅在卡内基和平基金会的同事、 fellow研究员马特·希恩认为，另一种途径是“并行推进人工智能安全”，即两国“并行前进，各自在推进技术发展的同时，也采取行动降低风险——因为他们相信这对自身国家安全至关重要，无论对方怎么做”。
存亡攸关的时刻
特朗普与习近平5月在北京会晤后，中国政府表示，两国领导人“同意就人工智能问题开展两国政府间对话”。
特朗普总统承认曾与习近平讨论过为人工智能发展“设置护栏”的想法，但在对此次峰会的分析中，战略与国际研究中心表示“几乎没有迹象表明在核心争端上取得了有意义的进展”，并指出白宫并未证实双方就该议题开启了任何新的对话。
特朗普总统定于9月24日在华盛顿再次与习近平会晤，预计双方将讨论人工智能安全问题。
不过，尽管特朗普总统对这些担忧予以否认，但许多分析师仍抱有一丝微弱的希望：对无节制的人工智能发展日益加剧的担忧，将推动两国领导人在一定程度上支持协调合作。
“一边是如此重大的担忧，”王蕾娅在谈及人工智能带来的生存危机时说，“另一边他们却在为这件事争执不休——而这会将我们推下悬崖。”
“我们这一代人做出的选择，将决定人类文明能否存续以及如何存续和发展，”易征补充道。
U.S. and China must get over mutual distrust on AI development before it’s too late, analysts warn
September 16, 2026 / 12:04 PM EDT / CBS News
As fears grow that the race for AI supremacy could be hurtling us all toward a grim fate not unlike the apocalyptic plot line of the Terminator movies, analysts tell CBS News that mutual distrust between the U.S. and China, as much as the technology itself, could lead humanity “off a cliff.”
The long-term fear among experts like the Anthropic researcher whose dire prediction stole headlines a week ago, is that superintelligence — a theoretical AI agent smarter than any person — could start to see humans as rivals for essential resources, endangering our survival.
The bosses of major U.S. AI firms OpenAI and Anthropic have now called for a slowing of AI development, to “pace the frontier.”
China, the other major hub of work to push that frontier alongside Silicon Valley, has had its own warnings. A team of U.S. researchers last week used AI to expose a flaw in China’s “app for everything” WeChat, which is used by more than 800 million people in the country.
Many Chinese firms agree with the growing consensus that AI must be regulated, according to Leia Wang, non-resident scholar at the Carnegie Endowment for Peace, who mentioned the Artificial Intelligence Laboratory in Shanghai as an example of a “highly reputable body” working on AI safety.
Yi Zeng, the dean of the Beijing Institute of AI Safety and Governance, wrote in a research paper shared with CBS News that “the catastrophic risks posed by AI, and the hidden danger of human extinction, are growing by the day.”
Britain’s King Charles III is to convene a meeting in Scotland on Thursday of industry leaders to discuss “whether a shared set of principles can be developed” to guide the future development and use of AI.
Nvidia CEO Jensen Huang, whose firm is among those attending the conference, said in April that it was essential that U.S. and Chinese researchers “are actually talking.”
This bilateral coordination by industry leaders “is glaringly missing because of our current attitude about China as an adversary,” he said.
Distrust, and “Chinese killer robots”
A group of Democratic U.S. lawmakers have called on House Speaker Mike Johnson to delay the upcoming recess to get legislation passed on AI safeguards.
Johnson said Sunday, however, that rushing regulations into law would mean the U.S. “will lose the race to China, and that is a threat to every single American.”
Republican Senator Ted Cruz said last week that “if there are gonna be killer robots, I’d rather they be American killer robots than Chinese killer robots.”
https://www.cbsnews.com/video/highlights-from-chinas-world-humanoid-games-and-what-it-means-for-the-future/
On Monday, President Trump dismissed warnings of an AI threat as a “sick conspiracy,” adding in his Truth Social post that “the only one that is happy about it is China. WHOEVER WINS AI, WINS!”
“It’s not that the U.S. government doesn’t think AI is risky,” Wang told CBS News, but that when it comes to U.S. leaders’ “China-facing AI thoughts, the number one concern is technological competition.”
China’s leaders have made some overtures toward collaboration on AI safety.
In June, President Xi Jinping explicitly warned of the dangers of a “loss of control” of the technology. A month later he said countries must work together to “ensure that AI is always under human control.”
But like many senior U.S. officials, China’s government has rejected calls to slow the pace of development, viewing the concept as an effort by the U.S. “to assert its technological dominance,” said Wang.
After Anthropic CEO Dario Amodei said the U.S. and its global partners should collaborate “to keep democracies’ AI lead over autocracies as large as possible, to give us the breathing room we need in order to pace effectively,” China’s Foreign Ministry said Monday that “narratives of threat … serve only to disrupt the process of global AI governance.”
Competition and mutual suspicion between the world’s two biggest economies may come as little surprise, but Wang told CBS News that reducing the “AI safety conversation into a technological competition conversation” was “the most unhelpful framing,” given the risks involved.
In a February essay entitled “Who will save the world from a U.S.-China AI arms race?”, National University of Singapore lecturer Alex Capri, an expert in trade and technology, wrote that the two countries “face a prisoner’s dilemma … both would be safer with restraint, yet each accelerates development to avoid falling victim to the other.”
Ways forward
“A coordinated slowdown or grand international agreement might not be the right frame to engage Chinese firms and the government,” Kwan Yee Ng, Head of International Governance at Concordia AI, an AI safety consultancy based in Beijing, told CBS News.
She added, however, that “there are many things the two countries can work on.”
That “low-hanging fruit,” according to Ng, includes sharing cybersecurity best practices, coordinating to avoid any misuse of AI by terrorist groups or agents seeking to cause economic turmoil, and opening up low-level government talks to share granular technical information.
Matt Sheehan, a colleague and fellow researcher of Wang’s at the Carnegie Endowment, argues that another approach is “AI safety in parallel,” which would see the two countries “move forward on parallel tracks, each advancing the technology and also acting to reduce the risks because they believe it is essential to their own national security — regardless of what the other does.”
An existential moment
China’s government said after a May meeting in Beijing between Mr. Trump and Xi that the leaders had “agreed to hold dialogue between the two governments” on AI.
President Trump acknowledged discussing the notion of “guardrails” on AI development with Xi, but in an analysis of the summit, the Center for Strategic and International Studies said “there was little indication of meaningful progress on the core disputes,” and it noted that the White House did not confirm any new dialogue with Beijing on the topic.
Mr. Trump is due to meet with Xi again in Washington on Sept. 24, and they are expected to discuss AI security.
President Trump and Chinese President Xi Jinping tour Zhongnanhai Garden on May 15, 2026, in Beijing, China. Evan Vucci-Pool/Getty/China Pool
The hope among many analysts, however faint, is that the growing urgency of concern over unbridled AI advancement, despite Mr. Trump’s rejection of the worries, will push the leaders to back coordination on some level.
“It’s like, here’s the big concern over here,” said Wang, referring to the existential AI fears, “and then they’re fighting about this thing — and it’s gonna drive us off a cliff.”
“The choices our generation makes will determine whether and how human civilization survives and develops,” added Zeng.