2026-02-02T10:00:46.898Z / CNN
持续数月的人工智能未来走向权力斗争于11月蔓延至副总统JD·万斯的办公室，当时唐纳德·特朗普总统的两名盟友进行了面对面的坦诚对话。
白宫人工智能沙皇大卫·萨克斯（David Sacks）在2025年试图将相关措辞纳入必须通过的联邦资金法案，这些法案本会抹去各州的人工智能法规，并让国会对这项强大技术的新监督权限受限。但长期以来对特朗普新科技盟友持怀疑态度的特朗普法律顾问迈克·戴维斯（Mike Davis）两次组织保守派活动人士和议员阻止了这一企图。与此同时，特朗普对这一优先事项缺乏进展公开表示不满。
据两名了解此次会议的人士透露，在万斯办公室，以好斗风格著称的戴维斯指责萨克斯试图绕过国会，在缺乏足够保障措施的情况下将人工智能强加给美国。萨克斯则反驳称，他只是在执行特朗普释放人工智能繁荣的意愿，而戴维斯在妨碍这一进程。
Vance最终鼓励萨克斯与戴维斯合作。几周后，特朗普签署了一项行政命令，部分由两人共同制定，旨在阻止各州执行自己的人工智能法规，并指示其政府与国会合作制定人工智能“单一国家框架”。该命令预计将面临法律挑战。
这一事件暴露出特朗普联盟内部在如何激进地释放这项正迅速重塑社会和经济的技术方面日益扩大的裂痕。一方是影响力日益增强的科技领袖及其盟友；另一方是担心工作被取代的工人阶级选民、担忧儿童安全的文化保守派，以及对科技行业深怀疑虑的MAGA忠实支持者。
尽管这一不稳定联盟为特朗普带来了短期胜利，但人工智能之争才刚刚开始——国会可能成为下一个战场。为此，科技公司已聘请数百名说客，向国会竞选活动捐赠数百万美元，并在中期选举前囤积资金支持亲AI的超级政治行动委员会。反对者也在准备动员。
“我们将拼命抗争，”特朗普前顾问、主要科技批评者史蒂夫·班农在总统签署行政命令后的播客中表示，“所以不要认为有人会感到满意。”
‘沙神’vs‘公民自由’
特朗普已迅速采取行动推动一项技术，从某些指标来看，该技术在他重返办公室第一年大部分时间里支撑了美国经济。随着萨克斯担任白宫人工智能和加密货币特别顾问，特朗普去年夏天制定了一个框架以快速推进人工智能项目。为在与中国的人工智能竞赛中占据优势，其政府还收购了芯片制造商英特尔公司10%的股份，并对外国芯片征收25%的关税。
但特朗普明确表示希望走得更远，承诺为科技公司提供它们渴望的监管自由。许多这些公司是其政治运作和新白宫宴会厅的最大金融支持者。
“我们将有大量投资流入，但如果他们必须获得50个不同州的50项不同批准，你可以忘了它，因为这是不可能做到的，”特朗普在签署行政命令时表示。
对这些努力的反对来自特朗普自己阵营中的有影响力人士。班农将其有影响力的“战争室”播客置于日益增长的反科技民粹主义浪潮的前沿，戴维斯是常客。
同样作为AI怀疑论者和“战争室”撰稿人的乔·艾伦（Joe Allen）走遍全国，试图敦促保守派听众抵制科技CEO及其将技术强加给人类的计划。
“他们最终的目标是建造‘沙神’，”艾伦说，“我担心世界上有足够多轻信的人，无论这些研究项目产出什么，都会被奉为神明。”
随着选举年临近，政府对AI的支持所带来的风险开始显现。皮尤研究中心9月的一项民调显示，一半美国人表示对AI与生活的日益融合更感担忧而非兴奋，而只有10%的人感到更兴奋而非担忧。
在各政治派别的社区中，地方领导人正响应公众压力，阻止或减缓AI项目，尤其是数据中心。越来越多的候选人将不断上涨的水电费归咎于耗能巨大的AI公司。
萨克斯认为，保守派对AI的“强烈”反感源于大流行病期间对大型科技公司的敌意以及对社交媒体的长期恐惧。他表示这是错误的。
“我认为，对公民自由担忧的右翼人士不应该希望政府在AI中扮演这种过度干预的角色，”他在播客中说。
但一位为科技客户提供政治策略建议的共和党人告诉CNN，AI公司应该担心日益增长的反对情绪，因为“从长远来看，特朗普可能会面临政治逆风并放弃AI”。
“这应该是行业真正的担忧，”该人士要求匿名以畅所欲言，“我认为这就是为什么支持者们大量讨论输给中国的国家安全风险。他们试图将特朗普逼入绝境。”
一些共和党人已开始与特朗普对AI的全面支持产生分歧。
当禁止州级AI法规的条款首次出现在一项包含特朗普大部分立法议程的国会预算协调法案中时，17位共和党州长联名致信国会要求将其从法案中删除。在佛罗里达州，州长罗恩·德桑蒂斯推动该州为AI制定自身保障措施，坚定宣称：“我们不会放弃任何权利。”与此同时，参议员乔希·霍利举行听证会，指责AI高管未能保护儿童。
随着更多共和党人将反科技立场融入政治品牌，这种反对情绪不仅给特朗普带来挑战，也给其最可能的继任者万斯带来压力。副总统长期以来试图在其与硅谷的深厚联系和民粹主义根源之间走钢丝——这一紧张关系在他办公室的戴维斯-萨克斯会议上显露无遗。
“美国掌权的途径是通过反科技寡头之门，”班农在被问及这一新兴分歧时告诉CNN，“你必须强硬、一致且真诚。”
即使在特朗普的AI支持者中，也越来越意识到公众情绪正迅速转向反对这项技术。12月，萨克斯和他“All-In”播客的联合主持人在与直言不讳的AI批评者、保守派评论员塔克·卡尔森对话时承认，该行业对日益增长的公众恐惧和批评反应迟缓。
卡尔森就能源消耗、工作岗位流失以及“这种技术可能完全失控并吞噬我们”等问题向主持人施压，同时嘲笑该技术推出过程中的不平衡。
“谁负责这个营销？”卡尔森问道。
“我不知道，”萨克斯回答，“是我吗？”
快速变化的情绪迫使白宫改变策略。
与11月泄露的草案相比，特朗普签署的最终行政命令明显有所缩减。
与该版本不同，最终版本称其政府“必须与国会合作”制定国家AI标准。该命令也不适用于各州对未成年人的保护或数据中心的监管，这是戴维斯长期以来要求的两项关键例外。
“我们正在积极参与并推动这一进程，”特朗普签署命令后戴维斯在班农的播客中表示。
萨克斯未回应置评请求。戴维斯也拒绝就万斯办公室的会议置评，告诉CNN：“我不会谈论我与白宫官员的私人讨论。”
“但大卫·萨克斯是个好人，他正本着诚意与我合作，为特朗普总统制定最佳AI政策，”他补充道。
在国会山，立法者首次有了塑造AI未来并回应选民对该技术日益增长的反对的真正动力。一名了解谈判情况的共和党参议院工作人员表示，在2025年大部分时间里，国会实际上被边缘化，而白宫内部派系辩论给予萨克斯和科技高管的自由度。但行政命令“应该是立法者采取行动所需的‘踢打’”。
与戴维斯密切合作的田纳西州共和党议员玛莎·布莱克本预计将在未来几周内提出新的国家AI规则。不过，特朗普已任命得克萨斯州参议员泰德·克鲁兹（科技行业盟友）牵头立法工作。
任何立法可能都需要民主党支持才能在参议院通过，但少数党尚未阐明自己的AI监管策略。民主党参议员此前曾联合阻止和解法案中的AI条款，但对AI潜力持乐观态度和希望严格监管该技术的议员之间仍存在广泛分歧。
参议院协议还必须通过众议院和多数党领袖史蒂夫·斯卡利斯（Steve Scalise），他去年曾试图在《国防授权法》中加入对州AI法律的全面禁令，后被戴维斯、布莱克本等人动员反对。
多次失败后，AI倡导者悄悄敦促萨克斯和斯卡利斯放弃这种方法。
去年年底，支持特朗普的超级政治行动委员会“Building America’s Future”发布了来自总统喜爱的民调机构的调查数据，显示美国人希望国会制定AI政策，而非各州拼凑的规则。但数据还显示，选民压倒性支持保护儿童免受AI危害的立法。
知情人士表示，这项民调旨在为斯卡利斯和其他国会亲AI共和党人指明潜在前进道路——这一路径可能得到大型科技公司支持，同时解决共和党亲家庭派系对AI的担忧。
一名与反对州级监管的团体合作的共和党人将这种政治局面称为共和党人的“雷区”。
“我们代表工人，如果我们不关注对就业的影响，毫无疑问，这将带来政治代价。如果我们不关注保护儿童，毫无疑问也会有政治代价，”该人士表示，“我认为总统意识到了这一点。”
随着2026年的斗争升温，AI行业主要参与者已更熟悉在华盛顿争取影响力的方式。
OpenAI联合创始人兼总裁格雷格·布罗克曼（Greg Brockman）去年向支持特朗普的超级政治行动委员会MAGA Inc.捐赠了2500万美元。另一个名为“Leading the Future”的超级政治行动委员会由行业利益支持，已积累约1亿美元用于针对反AI候选人。
但倡导者承认，这只能起到有限作用，可能需要改变AI叙事以对抗强烈反对。
“很多很多美国人害怕AI，而且不理解它，”今年发起AI基础设施联盟（AI Infrastructure Coalition）支持该行业的前亚利桑那州参议员克里斯滕·西尼玛（Kyrsten Sinema）表示，“AI公司在帮助人们看到AI在日常生活中的作用方面做得不够出色。这个故事需要被讲述。”
埃隆·马斯克最近认为，未来二十年内，AI自动化和机器人技术将使工作成为可选，人类将拥有所需一切，大多数人将依靠全民基本收入生活。天使投资人、萨克斯“All-In”播客联合主持人杰森·卡拉卡尼斯（Jason Calacanis）呼吁开展“曼哈顿计划”，全国可以围绕这一计划团结起来，包括“10个拥有1000万套新住房的新城，以及全民免费医疗和职业学校免费教育”。
“这就是解决问题的方法，”他表示，“这是没有人在做的事。”
但这些幻想场景几乎没有安抚最强烈的反对者。艾伦称卡拉卡尼斯的提议是“如何取代所有人并安抚民众”的计划。
卡尔森则将这种乌托邦式愿景斥为“我最反感的AI讨论内容”。
CNN的哈达斯·戈尔德（Hadas Gold）和克里斯汀·霍姆斯（Kristen Holmes）对本文有贡献。
Trump’s AI push exposes a divide in the MAGA movement
2026-02-02T10:00:46.898Z / CNN
A monthslong power struggle over the future of artificial intelligence spilled into Vice President JD Vance’s office in November, when two of President Donald Trump’s allies met face-to-face for a frank conversation.
David Sacks, the White House AI czar, had spent 2025 trying to tuck language into must-pass federal funding bills that would have wiped away state AI regulations and left Congress with limited new oversight of the powerful technology. But Mike Davis, a longtime Trump legal adviser skeptical of the president’s new tech allies, twice helped rally conservative activists and lawmakers to stop it. Trump, meanwhile, had grown publicly frustrated at the lack of progress on one of his top priorities.
In Vance’s office, Davis, known for his combative style, accused Sacks of trying to run over Congress and impose artificial intelligence on the country without sufficient safeguards, according to two people with knowledge of the meeting. Sacks countered that he was simply carrying out Trump’s desire to unleash an AI boom, and Davis was getting in the way.
Ad Feedback
Vance ultimately encouraged Sacks to work with Davis. A few weeks later, Trump signed an executive order, shaped in part by both men, that aims to block states from enforcing their own artificial intelligence regulations and directs his administration to team up with Congress to create a “single national framework” for AI. The order is widely expected to face legal challenges.
The episode laid bare a growing fault line within Trump’s coalition over how aggressively to unleash a technology that is rapidly reshaping society and the economy. On one side are increasingly influential tech leaders and their allies. On the other are working-class voters fearful of job disruption; cultural conservatives worried about child safety; and MAGA loyalists who view the industry with deep suspicion.
While the uneasy alliance delivered Trump a short-term victory, the battle over AI is just beginning — and Congress may be the next front. In anticipation, tech companies have hired hundreds of lobbyists and donated millions of dollars to congressional campaigns, and they are stockpiling cash in AI-friendly super PACs ahead of the midterms. Opponents are also preparing to mobilize.
“We’re going to fight like hell,” Steve Bannon, a former Trump adviser and a leading tech critic, said on his podcast after the president signed the executive order. “So don’t think that anybody is placated.”
‘Sand gods’ vs. ‘civil liberty’
Trump has already moved quickly to boost a technology that, by some measures, helped prop up the US economy through much of his first year back in office. With Sacks working as a special White House adviser on artificial intelligence and cryptocurrencies, Trump laid out a framework last summer to fast-track AI projects. To gain an edge in the AI race against China, his administration also took a 10% stake in the chipmaker Intel Corporation and imposed 25% tariffs on foreign chips.
But Trump has made clear he wants to go further, promising to deliver for tech companies the regulatory freedom they crave. Many of those companies are among the largest financial backers to his political operation and his new White House ballroom.
“We have the big investment coming, but if they had to get 50 different approvals from 50 different states, you can forget it, because it’s not possible to do,” Trump said when he signed the executive order.
Opposition to these efforts has emerged from influential voices within Trump’s own movement. Bannon has positioned his influential “War Room” podcast at the vanguard of a growing wave of anti-tech populism. Davis is a regular guest.
So is Joe Allen, a leading AI skeptic and “War Room” contributor who has traveled the country trying to urge conservative audiences to push back against tech CEOs and their plans to force their technology on humanity.
“They are ultimately aiming toward building sand gods,” Allen said. “And I fear there are enough credulous people in the world that whatever comes out of these research projects will be worshipped as a god.”
The risks of the administration’s embrace of AI are beginning to crystalize heading into an election year. Half of Americans say they are more concerned than excited about AI’s increasing intersection with their lives, according to a Pew Research Center poll from September, while just 10% feel more excited than concerned.
In communities of all political stripes, local leaders are responding to public pressure to block or slow AI projects, especially data centers. Increasingly, candidates for office are blaming rising utility bills on energy-hungry AI companies.
Sacks has argued the “very visceral” conservative disdain for AI stems from hostility toward Big Tech dating to the pandemic and from lasting fears about social media. He says that’s misguided.
“I don’t think that people on the right who are concerned about civil liberty should want the government to play this super-intrusive role in AI,” he said on his podcast.
But one Republican who advises tech clients on political strategy told CNN that AI companies should be concerned by growing backlash because “there’s a potential in the long run for Trump to see political headwinds and walk away from AI.”
“It should be a genuine concern of the industry,” said the person, who asked not to be named in order to speak freely. “I think that’s why there’s so much discussion from proponents about the national security risks of losing the AI race to China. They’re trying to box Trump into a corner.”
Some Republicans are already breaking from Trump’s full embrace of AI.
When language banning state-level AI regulations first surfaced in a congressional budget reconciliation package that carried much of Trump’s legislative agenda, 17 Republican governors sent a letter calling on Congress to strip it from the bill. In Florida, Gov. Ron DeSantis has pushed ahead with proposals for his state to put its own safeguards around AI, defiantly declaring, “We’re not going to give up any rights.” Sen. Josh Hawley, meanwhile, has held hearings to admonish AI executives for failing to protect children.
As more Republicans stitch anti-tech stances into their political brands, the backlash is creating challenges not only for Trump, but also for Vance, his most likely successor. The vice president has long tried to straddle a line between his deep ties to Silicon Valley and his populist roots — a tension on display in the Davis-Sacks meeting in his office.
“The path to power in America is through the anti-tech oligarch gate,” Bannon said to CNN when asked about this emerging divide. “You have to be hard, consistent and authentic.”
Even among Trump’s AI cheerleaders, there is a growing realization that public sentiment is quickly shifting against the technology. In December, Sacks and the co-hosts on his “All-In” podcast acknowledged that the industry had been slow to respond to mounting public fears and criticism during a conversation with conservative commentator Tucker Carlson, an outspoken AI critic.
Carlson pressed the hosts on concerns ranging from energy consumption to job disruption and “the potential this gets completely away from us and eats us.” But he also mocked the technology’s uneven public rollout.
“Who’s in charge of the marketing for this?” Carlson asked.
“I don’t know,” Sacks replied. “Me?”
The fast-changing sentiments have forced a shift in tactics from the White House.
The final executive order Trump signed was noticeably scaled back compared with a draft that leaked in November.
Unlike that version, the final copy said his administration “must act with the Congress” on a national AI standard. The order also won’t apply to state protections for minors or regulations of data centers, two critical carve-outs that Davis had long demanded.
“We’re very much at the table and driving this process,” Davis said on Bannon’s podcast after Trump signed the order.
Sacks did not respond to a request for comment. Davis also declined to address the meeting in Vance’s office, telling CNN: “I am not going to talk about my private discussions with White House officials.”
“But David Sacks is a good man who is working with me in good faith for the best AI policy for President Trump,” he added.
On Capitol Hill, there is for the first time real momentum for lawmakers to shape the future of AI and respond to growing voter backlash over the technology. For much of 2025, Congress was effectively sidelined while factions within the White House debated how much latitude to give Sacks and tech executives, said a Republican Senate staffer with knowledge of the negotiations. But the executive order “should be the kick in the pants lawmakers need to act,” the staffer said.
Sen. Marsha Blackburn, a Tennessee Republican who has worked closely with Davis, is expected to introduce new national AI rules in the coming weeks. Trump, though, has tapped Texas Sen. Ted Cruz, an ally of the tech industry, to take the lead on legislation.
Any legislation would likely need Democratic support to pass the Senate, and the minority party has yet to articulate its own strategy for regulating AI. Democratic senators previously banded together to block AI language from the reconciliation package, but a broader split remains between lawmakers optimistic about AI’s potential and those eager to crack down on the technology.
A Senate deal must also get through the House and Majority Leader Steve Scalise, who tried last year to tack on to the National Defense Authorization Act a blanket ban on state AI laws before Davis, Blackburn and others mobilized opposition.
After multiple defeats, AI advocates have quietly urged Sacks and Scalise to move off that approach.
Late last year, the pro-Trump super PAC Building America’s Future released survey data from the president’s favorite polling firm that suggested Americans want Congress to set AI policy, rather than a patchwork of states. But their data also showed voters overwhelmingly supported legislation protecting children from AI’s more problematic powers.
The polling was intended to signal a potential path forward for Scalise and other aligned Republicans in Congress — one that could be supported by Big Tech but also address some looming child safety concerns that those in the pro-family wing of the GOP have with AI, according to a person with knowledge of the strategy.
One Republican working with groups advocating against state regulations called the political landscape a “minefield” for the GOP.
“We represent working people, and if we’re not sensitive towards the impact on jobs, no question, there’s going to be political cost to that. If we’re not sensitive to protecting children, no question” there will be a political cost, the person said. “I think the president’s aware of that.”
As the 2026 fight heats up, major players in the AI industry have become more versed in the ways of winning influence in Washington.
Greg Brockman, the co-founder and president of OpenAI, a company at the forefront of the artificial intelligence boom, gave $25 million to the pro Trump super PAC MAGA Inc. last year. Another super PAC, called Leading the Future, is backed by industry interests and has amassed about $100 million to target anti-AI candidates.
But that will only go so far, advocates acknowledge, and a change in the AI narrative might be needed to combat a vocal opposition.
“Lots and lots of Americans are scared of AI and don’t understand it,” said former Arizona Sen. Kyrsten Sinema, who launched the AI Infrastructure Coalition this year to advocate for the industry. “AI companies haven’t done the most excellent job of helping people see AI in their daily lives. That story needs to be told.”
Elon Musk recently argued that within two decades, AI automation and robotics will eventually make work optional, humans will have everything they need and most people will live off a universal income. Jason Calacanis, an angel investor and Sacks’ “All-In” co-host, called for “a Manhattan Project” that the country can rally around. It would include “10 new cities with 10 million new homes and free health care for everybody and free education for trade schools,” he said.
“That’s what solves the problem,” he said. “That’s what nobody’s doing.”
But those fantastical scenarios have hardly appeased the loudest dissenters. Allen called Calacanis’ proposal a plan for “how to replace everyone and keep the population placated.”
Carlson, for his part, dismissed such utopian visions as “the thing that offended me most about the AI conversation.”
CNN’s Hadas Gold and Kristen Holmes contributed to this report.