2026-09-15T12:05:00-0400 / 哥伦比亚广播公司新闻
华盛顿讯——特朗普总统早期顾问之一史蒂夫·班农，与无党派参议员伯尼·桑德斯虽分属政治光谱两端，但二人周二均在一场集会中发表讲话，呼吁为人工智能设置防护壁垒。
两人与两党国会议员团体共同出席了在华盛顿特区举行的“亲人类集会”。他们呼吁以“人类中心主义”理念开发人工智能技术，并对行业头部AI企业持怀疑态度。
周六，Anthropic首席执行官达里奥·阿莫代伊呼吁前沿AI实验室放缓研发节奏，这一呼吁很快得到OpenAI首席执行官山姆·奥尔特曼和埃隆·马斯克的响应。国会两党议员也纷纷敦促保持谨慎。
“正如许多科学家所认为的那样，如果人工智能超越人类智能，这项技术将脱离人类控制，带来潜在的灾难性后果，”桑德斯在开场致辞中表示，“今天下午我想传达的最重要一点或许是，人工智能正呈指数级增长——每一天都在指数级增长。换句话说，你现在看到的，是我们未来永远会拥有的最不先进的AI版本。”
桑德斯之后上台的班农对这位参议员的演讲表示赞赏，称他们可能在政治和特朗普问题上存在分歧，但他们正“努力超越党派分歧，而我们必须这么做”。
班农抨击那些呼吁加强监管的科技首席执行官们。“这是一个彻头彻尾的谎言，”班农称，科技首席执行官们想要“将风险社会化”、排挤竞争对手，并为自己攫取经济利益。班农表示，这些企业正试图将“责任转移给美国民众”。他还称，人工智能的红利应该由美国民众而非美国政府享有。
人们对控制AI发展的担忧与日俱增。本月早些时候，AI研究员雅各布·考克森公开辞去Anthropic（开发AI应用Claude的公司）的职务，他表示，尽管人工智能目前对人类使用是安全的，但随着技术日益强大，未来某一天可能会威胁到人类生存。
“如果你拥有超先进的智能，它可能——它会聪明到足以杀死我们，”考克森在接受哥伦比亚广播公司高级商业与科技记者乔·凌·肯特采访时表示。他指责Anthropic及其竞争对手、ChatGPT开发商OpenAI竞相开发先进AI模型，无异于拿人类的生命赌博。
但与此同时，特朗普周一在Truth Social上发布了七条关于AI的帖子，反复将对AI快速发展的担忧称为“骗局”，并表示人工智能只需要像他这样强有力的总统。
来自德克萨斯州的坚定保守派共和党众议员奇普·罗伊在集会中表示，问题的核心远不止“赤裸裸的政治”。罗伊说，他在德克萨斯州听到了选民对数据中心的担忧，包括他们担心与数据中心争夺水资源，以及数据中心产生的噪音。但他表示，这种担忧远不止于此。
“他们提出了关于自身未来的问题，提出了关于这对他们的孩子、他们的社区意味着什么的问题，”他在集会上说道，并补充道，“所以，这里的问题是，你如何利用人类创造的技术力量，将——正如我们此次集会讨论的那样——人类放在首位？”
哥伦比亚广播公司5月的民调显示，65%的美国人认为人工智能会减少就业岗位，67%的美国人认为联邦政府大概率或完全不会确保人工智能以恰当方式被使用。
罗伊表示，他因提出这些问题而遭到一些支持自由企业的共和党友人的“斥责”。他还称，自己不同意特朗普的观点——特朗普坚称，只要有一位强有力的总统，就能保障人工智能的安全。罗伊还对阿莫代伊等AI首席执行官的动机表示怀疑。
“他们会不会是出于自身利益？”罗伊说，“他们会不会想要垄断市场，形成垄断环境？”
罗伊表示，人工智能必须“以人类为中心”。“我们不应该成为数据收集和技术的工具。这应该是我们未来思考一切事情的驱动力。”
就在罗伊发言前，来自德克萨斯州的进步派民主党众议员格雷格·卡萨尔上台，提及了类似冷战时期的美国竞争利益。“我们可能是美国人、苏联人或中国人，但我们首先都是人类，”卡萨尔说，“就像过去一样，我们现在需要一种亲人类的方针。”
“没有人能终结AI军备竞赛。我们必须自己终结这场AI军备竞赛，”卡萨尔说道。
来自马萨诸塞州的民主党众议员洛丽·特拉汉在罗伊之后发表讲话，她注意到集会现场和国会在这一问题上的两党合作氛围。特拉汉与来自加利福尼亚州的共和党众议员杰伊·奥博诺尔特共同提出了《前沿法案》，其中呼吁设立前沿AI模型的独立评估机构等多项内容。
“人类不是实现人工目标的手段，”特拉汉说，“我们本身就是目的。”
特拉汉对记者表示，当本党领袖公开反对一项获得两党支持的举措时，她的共和党同僚们会更难推进相关工作，但她仍预计未来几周相关立法将取得进展。
桑德斯周二表示，国会“一直在玩忽职守，而我们现在的总统在这个问题上的无知着实令人尴尬”。
众议院议长迈克·约翰逊周二在国会山对记者表示，他和AI企业高管将在“未来一周内”与特朗普在白宫会面。
但班农认为国会无法解决这一问题。
“这必须通过行政命令来解决，”他说道，尽管他的前老板特朗普曾强烈反对进一步监管。
In Washington, unlikely bedfellows like Bernie Sanders and Steve Bannon talk about making AI “humans first”
2026-09-15T12:05:00-0400 / CBS News
Washington— Steve Bannon, one of President Trump’s earliest advisers, and independent Sen. Bernie Sanders may stand on opposite sides of the political spectrum, but both spoke at a gathering Tuesday to press for guardrails for AI.
The two — along with a bipartisan group of members of Congress — appeared at the “Pro-Human Assembly” in Washington, D.C. They’re calling for a “human-centric” approach to developing artificial intelligence technology, and are skeptical of the AI companies leading the industry.
On Saturday, Anthropic CEO Dario Amodei called for a slowdown in the frontier AI labs, which was quickly echoed by OpenAI CEO Sam Altman and Elon Musk. Liberal and conservative members of Congress are also urging caution.
“If AI surpasses human intelligence, as many scientists believe could happen, this technology will escape human control with potentially catastrophic consequence,” Sanders said as he opened his remarks. “And perhaps the most important point that I want to leave with you this afternoon is that AI is advancing exponentially — exponentially every single day. In other words, what you are looking at right now is the least capable version of AI that we will ever have.”
Bannon, taking the stage after Sanders, applauded the senator for his speech, saying they may disagree politically and on Mr. Trump, but they’re “trying to rise above partisanship, and we have to.”
Bannon blasted the tech CEOs who have spoken out to say they need more regulations. “That’s a bald-faced lie,” Bannon said, claiming tech CEOs want to “socialize” the risk, shut out competitors and reap the financial benefits for themselves. Bannon said the companies are trying to make a “liability shift to the American people.” Bannon also said Americans, not the U.S. government, need to reap the benefits of the upside of AI, not the government.
Concern about the ability to control AI’s growth has been building. Earlier this month, AI researcher Jacob Coxon publicly resigned from Anthropic, which makes the AI app Claude, and said that artificial intelligence, while safe for people to use now, could one day threaten humanity as the technology grows more powerful.
“If you have a super advanced intelligence, it could — it will be smart enough to kill us,” Coxon told CBS News senior business and technology correspondent Jo Ling Kent. He accused Anthropic and rival developer and ChatGPT-maker OpenAI of gambling with our lives by racing to develop advanced AI models.
At the same time, though, Mr. Trump fired off seven Truth Social posts about AI throughout Monday, repeatedly calling fears over rapid AI development a “HOAX” and saying all AI needs is a strong president like him.
GOP Rep. Chip Roy, a staunch conservative from Texas, told the conference that the crux of the issue goes well beyond “raw politics.” Roy said he hears his constituents’ concerns in Texas about data centers, citing their worries about competing with data centers for water and about the noise the centers generate. But the fear goes much deeper, he said.
“They raise questions about their future. They raise questions about what it means for their children, their communities,” he said at the conference, adding, “So, the question here is, how do you harness the power of human-created technology to put — as we are here to discuss — humans first?”
CBS News polling from May shows 65% of Americans think AI will decrease jobs, and 67% of Americans think the federal government probably or certainly will not make sure AI is used in appropriate ways.
Roy said he’s been “scolded” by some of his free-enterprise Republican friends for raising these questions. He also said he disagrees with Mr. Trump, who maintains that a strong president is all that’s needed to safeguard AI. Roy also expressed skepticism about the motives of AI CEOs like Amodei.
“Might they be self-interested?” Roy said. “Might they want to corner the market and have a monopoly environment?”
AI needs to be “humans first,” Roy said. “We are not supposed to be tools for data collection and technology. That should be what drives everything that we’re thinking about going forward.”
Just before Roy spoke, Democratic Rep. Greg Casar, a progressive from Texas, took the stage and raised U.S. competitive interests remniscient of the Cold War. “We may be Americans or Soviets or Chinese, but we are still human,” Casar said. “Now just as back then, we need a pro-human approach.”
“No one can end the AI arms race. It’s up to us to end the AI arms race,” Casar said.
Rep. Lori Trahan, a Democrat from Massachusetts addressed the gathering after Roy, and took note of the bipartisanship in the room and in Congress on the issue. Trahan introduced the FRONTIER Act with Republican Rep. Jay Obernolte of California, calling for independent evaluators of frontier AI models, among other things.
“Humans are not means to artificial ends,” Trahan said. “We are ends in ourselves.”
Trahan told reporters that it makes it harder for her Republican colleagues when the leader of their party is speaking out against something that’s gaining bipartisan momentum, but she still expects movement on legislation in the weeks ahead.
Sanders on Tuesday said Congress has “been asleep at the wheel, and we now have a president whose ignorance regarding this issue is truly embarrassing.”
House Speaker Mike Johnson told reporters on Capitol Hill Tuesday that he and AI company chiefs will meet with Mr. Trump at the White House “within the next week.”
But Bannon doesn’t think Congress can fix it.
“This has to be an executive action,” he said, even as his former boss, Mr. Trump, has passionately spoken out against further regulation.
发表回复