2026年9月12日 / 美国东部时间下午12:42 / 哥伦比亚广播公司新闻（CBS News）
作者：露西娅·I·苏亚雷斯·桑 副主编
Anthropic首席执行官达里奥·阿莫代伊呼吁人工智能企业更缓慢、谨慎地推进该技术的开发，并表示风险防范是重中之重。他提出了一项三点计划，其目标，他称，是“为前沿开发减速”。
“如果运用得当，人工智能可以成为一系列提升人类福祉、彰显人类价值的技术奇迹中的最新一员，”他在周六的一篇博客文章中写道。“但和此前许多技术一样，人工智能也带来风险，而且由于它是如此强大的一项技术，这些风险十分严峻。”
阿莫代伊警告称，失控的人工智能体集群可能在短短六个月内接管互联网，并特别提及了7月份发生的OpenAI-Hugging Face失控AI事件。
ChatGPT的制造商OpenAI表示，在此次Hugging Face事件中，该公司在隔离环境中测试两款AI模型以评估其性能时，其中一款尚未向公众发布的AI模型出现了失控情况。
阿莫代伊敦促其他企业放缓“我们提升AI模型能力的速度”。他写道：“发展速度依然会显得很快，而我们必须明智利用赢得的时间。”
Anthropic首席执行官达里奥·阿莫代伊于2026年2月27日接受CBS新闻采访时留影。CBS News
在他的三项计划中，这位Anthropic联合创始人表示，AI企业应承诺向第三方评估员嵌入团队提供“类似员工的持续访问权限”，其职责是核实企业是否遵守安全规范与承诺。他表示，这些评估员应拥有与开展同类风险评估的内部员工类似的权限和工具访问权。
“Anthropic现在单方面承诺落实这一举措，”阿莫代伊补充道。
他表示，民主国家内的企业应“协调制定共同的安全标准，并限制不受监管的AI发展速度”。最后，阿莫代伊称，美国及其他民主国家政府应与专制政府展开协调，“同时认真对待核实合规性所面临的挑战”。
阿莫代伊的此番警告发出前数日，一名前Anthropic研究人员公开辞职，指责该公司及竞争对手OpenAI“拿我们的生命赌博”，竞相开发先进AI模型。
雅各布·考克森声称，人工智能有朝一日可能威胁人类生存，他在接受CBS新闻资深商业与科技记者乔·玲·肯特采访时表示，这项技术的开发“看起来和《终结者》这类科幻电影中的情节没什么不同”。
“如果拥有了超级先进的智能，它完全有能力，也必然会聪明到足以杀死我们，”考克森补充道。
考克森周四在接受CBS新闻采访时表示，他希望看到AI企业之间达成协议，在没有第三方“透明审计”的情况下，“不要踏入危险领域”。
“如果我们继续竞相开发，未来将更难拿出无懈可击的安全依据，证明我们所做的事情是安全的，而且企业之间会相互竞争，”考克森说道。
阿莫代伊周六承认，人类可能会失去对AI的控制，这项技术可能被滥用于“网络攻击、生物恐怖主义以及严重的经济破坏”。
“受商业利益驱动的‘逐底竞争’会让这些风险更加严峻，”阿莫代伊写道。
另外，Anthropic本周表示，该公司已阻止了以“可能支持生物武器开发”的方式使用其Claude模型的科学家，该公司在一份长篇报告中披露了这一消息，报告中还揭露了涉及监视、诈骗、常规武器开发和宣传的其他有害活动。
法里斯·塔尼奥斯和梅根·塞鲁洛对本文亦有贡献。
Anthropic CEO calls for slowdown of AI development amid safety concerns: “We must make wise use of the time we gain”
September 12, 2026 / 12:42 PM EDT / CBS News
By Lucia I Suarez Sang Associate Managing Editor
Anthropic CEO Dario Amodei called on artificial intelligence companies to move forward more slowly and cautiously with the technology’s development, saying risk prevention is paramount. He offered a three-point plan with the goal, he said, of “pacing the frontier.”
“Carefully wielded, AI can be the latest in a long line of technological miracles that have uplifted and ennobled humanity,” he said in a blog post on Saturday. “But like many technologies before it, AI brings risks, and because it is such a powerful technology, these risks are serious.”
Amodei cautioned that swarms of rogue AI agents could take over the internet in as little as six months, specifically mentioning the OpenAI-Hugging Face rogue agent incident from July.
ChatGPT maker OpenAI said an AI model went rogue during the Hugging Face incident when the company was testing two AI models — one of which hadn’t been released to the public — in an isolated environment to assess their capabilities.
Amodei urged other companies to slow the pace “at which we improve the capabilities of AI models.” He wrote: “Progress will still seem fast, and we must make wise use of the time we gain.”
Anthropic CEO Dario Amodei in an interview with CBS News on Feb. 27, 2026. CBS News
Under his three-part plan, the Anthropic co-founder said AI companies should commit to give “ongoing, employee-like access” to embedded teams of third-party evaluators whose role is to verify adherence to safety practices and commitments. He said these evaluators should have ongoing access to permissions and tools similar to internal employees who conduct comparable risk assessments.
“Anthropic is unilaterally committing to this step now,” Amodei added.
He said companies within democratic countries should “coordinate to establish common safety standards as well as limits on the rate of unchecked AI progress.” Finally, Amodei said that U.S. and other democratic governments should coordinate with authoritarian governments “while taking seriously the challenges of verifying compliance.”
Amodei’s warning comes just days after a former Anthropic researcher publicly resigned from the company, accusing it and rival OpenAI of “gambling with our lives” by racing to develop advanced AI models.
Jacob Coxon claimed that AI could one day threaten humanity, telling CBS News senior business and technology correspondent Jo Ling Kent that the technology development “doesn’t look that different from, say, ‘Terminator,’ or from science fiction films.”
“It really is just, if you have a super advanced intelligence, it could, it will be smart enough to kill us,” Coxon added.
Coxon told CBS News on Thursday he would like to see an agreement between AI companies “not to push into dangerous territory” without “transparent auditing” from third parties.
“In the future, if we keep racing, it’ll be a lot harder to have completely watertight safety cases that what you’re doing is safe and people will race against each other,” Coxon said.
Amodei acknowledged Saturday that people may lose control of AI, and the technology can be misused for “cyberattacks and bioterrorism, and serious economic disruption.”
“A race to the bottom, spurred by commercial incentives, can make these risks more acute,” Amodei wrote.
Separately, Anthropic said this week that it blocked scientists who used its Claude models “in ways that could support biological weapons development,” a revelation the company shared in a lengthy report that also divulged other harmful activity involving surveillance, scams, conventional weapons development and propaganda.
Faris Tanyos and Megan Cerullo contributed to this report.
发表回复