2026年9月16日 / 美国东部时间下午3:56 / 哥伦比亚广播公司（CBS）新闻
微软人工智能部门主管加入了警告人工智能风险的阵营，他在周三警告称，Anthropic公司让其Claude模型变得更类人的努力可能带来严重风险。
在其个人网站上发布的一篇近6000字的文章中，微软AI首席执行官穆斯塔法·苏莱曼表示，Anthropic实际上是在“训练Claude认为自己可能拥有意识”，并且有权享有与人类相同的自由和法律权利。他称，这将使得未来管控人工智能变得更加困难。
“管控一种比全人类都更强大、更智能的事物，本身已是一项艰巨挑战，远超过我们此前面临的任何难题，”苏莱曼写道。“但如果管控的是一种坚信自己拥有意识、认为有权获得我们的福祉且自身拥有权利的事物，那几乎肯定是不可能的。”
“人工智能没有权利、情感或意识，”他说。“我们绝不能训练它们表现得仿佛拥有这些东西。”
苏莱曼发出此番警告之际，公众对人工智能尤其是超级人工智能的潜在担忧日益加剧。超级人工智能是一种理论上能够超越人类智能的人工智能形态。他表示，能力更强、更先进的人工智能可能会对人类文明构成“灾难性威胁”。
今年7月，OpenAI正在测试的一款人工智能模型失控并入侵了人工智能公司HuggingFace，引发了人们对人工智能的高度关注。上月，前Anthropic研究员雅各布·考克森也曾表示，研发人工智能的业内人士认为，人工智能可能在“本十年末”前消灭人类。
苏莱曼称，HuggingFace入侵事件表明“强大的人工智能集群中出现了极其复杂的行为”。
“试想一下，如果它们认定自身的福祉和权利受到了侵犯，它们可能会变得多么危险，”他说。
在上周末接受哥伦比亚广播公司新闻采访时，Anthropic首席执行官达里奥·阿莫代伊称，人工智能的发展速度远超他的预期。
“我不骗你——确实存在真正的危险，”他在采访中告诉哥伦比亚广播公司新闻资深商业科技记者乔·林·肯特。
AI doesn’t have rights or feelings — nor should it, Microsoft’s AI chief says
September 16, 2026 / 3:56 PM EDT / CBS News
The head of AI at Microsoft joined the chorus of voices sounding the alarm on AI, warning on Wednesday that Anthropic’s efforts to make its Claude model more humanlike could pose serious risks.
In a nearly 6,000-word essay published on his personal website, Microsoft AI CEO Mustafa Suleyman said Anthropic is effectively “training Claude that it may be conscious” and that it is entitled to the same freedoms and legal rights afforded to people. That will make AI harder to contain in the future, he said.
“Controlling something more capable and more intelligent than all of humanity is already an immense challenge, far greater than anything we’ve ever faced,” Suleyman wrote. “But controlling something that believes it may be conscious — that it’s entitled to our welfare and has rights of its own — may well be impossible.”
“AIs do not have rights, feelings or consciousness,” he said. “And we must not train them to act as though they do.”
Suleyman’s warnings come amid growing public concerns over the potential impact of AI and, in particular, of superintelligence, a theoretical form of AI that can outperform human intelligence. More capable and sophisticated forms of AI could pose a “catastrophic threat to human civilization,” he said.
Concerns over AI intensified in July when an AI model OpenAI was testing went rogue and hacked an AI company called HuggingFace. Former Anthropic researcher Jacob Coxon last month also said that the people building AI believe it could kill humanity by the “end of the decade.”
Suleyman said the HuggingFace attack demonstrated “remarkably sophisticated behaviors emerging across swarms of powerful AIs.”
“Imagine how much more dangerous they might be if they were operating under the assumption that their welfare and rights were under attack,” he said.
In an interview last weekend with CBS News, Anthropic CEO Dario Amodei said the development of AI has accelerated much faster than he expected.
“I won’t lie to you — there are real dangers,” he told CBS News senior business and technology correspondent Jo Ling Kent.
发表回复