Exclusive: Barack Obama’s hidden hand shaping Democrats’ AI debate

Jim Vondruska/Getty Images/File

Yes, the likely 2028 Democratic presidential candidates who’ve talked to Barack Obama keep saying when he asks them about AI, they’re paying attention and working up plans.

Obama doesn’t think what he’s seen so far is anywhere near enough — for addressing the scale of the issue, and for dealing with the politics. And he tells them that.

The former president isn’t alone: Democratic leaders across the spectrum, from former Federal Trade Commission Chair Lina Khan to former Commerce Secretary Gina Raimondo, are also demanding more from a coalescing field they say so far has generally been more reactive than meeting the moment. But none of them — including Obama — can agree yet quite what the plan should be, and some 2028 hopefuls are worried about getting on the wrong side of the AI industry’s massive campaign spending.

Still his party’s standard-setter 10 years out of the White House, Obama is regularly in touch with the rising generation of Democratic leaders, from when they reach out for advice to when he calls to congratulate them on victories. But on trips to California, in his office in Washington and over the phone, Obama has also been talking with leaders of all the major artificial intelligence companies — including Sam Altman of OpenAI, as well as executives at Google and Anthropic — and a range of outside experts and academics along with informal advisers, including Jason Goldman, the former Twitter executive who ran his Office of Digital Strategy, according to people familiar with the conversations.

He’s deep into the reading, favoring articles by Ezra Klein, Derek Thompson, Charlie Warzel and Casey Newton, and serving as an informal bridge between AI founders updating him and asking advice for how to navigate Washington and the next generation of rising politicians turning to him as they feel through what to do.

President Donald Trump hosted several AI leaders at the White House on Tuesday, saying they had signed a new commitment aimed at better “self-policing” their companies’ AI models. (Trump also said he would soon sign an executive order renaming “artificial intelligence” as “super intelligence.”) Most Republicans in Washington have taken their cues from him, pushing back against broad federal regulation of AI.

Julia Demaree Nikhinson/AP

The real reason why Donald Trump is so furiously against AI guardrails

Democrats eyeing 2028, meanwhile, have been releasing speeches and commissions and plans that — multiple party operatives concede — are fledging attempts to show they’re doing something. But they still largely don’t understand where the technology is now or how much it could change in two years.

Advisers to several potential candidates admit privately to CNN they’re so far avoiding doing much of anything they could get tagged with —caught between coming off too captive of the corporations or too combative, insufficiently populist or over-regulatory. They’re also cognizant that speaking out could make the companies’ massively funded super PACs target them in a crowded primary.

Obama doesn’t talk about humanity ending, staying more focused on the fallout in job displacement, national security and the effect especially on children from the mushrooming of disinformation. For months — long before the recent rash of warnings from AI executives — he has been telling every Democrat who comes to see him or who he’s been on the phone with even for congratulatory calls that they need to have a strong sense of what to do, for both policy and political reasons, according to multiple people familiar with the conversations.

“Given the scale of disruption we are likely to face, President Obama believes it’s important to have a grounded and informed conversation about how this technology is developed and introduced,” his senior adviser Eric Schultz told CNN. “Its effects on our economy and our democracy will depend on the choices we make. That means understanding the enormous opportunities AI presents while being clear-eyed about the risks and how best to mitigate them.”

Over 10 years, Obama has gone from joking in an interview about AI — “When you see it about to happen, you gotta yank that electricity out of the wall” — to a grim view of how unprepared the country is for the consequences of what is arriving .

Earlier this month during an appearance at Colgate University , he said, “in terms of how do we cure cancer or get better energy, you can do that without having agentic AI and having it just roaming free in the internet. The reason you are doing that is because you have to market a product that people will pay money for.”

Kent Nishimura/AFP/Getty Images

The rise of artificial intelligence and its coming impact would still be really hard, Obama has said in multiple conversations relayed to CNN, even if he felt the country had a competent administration in charge. That he has little confidence in Trump — who last week announced at the United Nations he’d rather call it “super intelligence” and on Tuesday met with several AI founders himself as he keeps to a laissez-faire approach — Obama tends to add, makes it infinitely harder.

Interested in changing technology going back to his White House days and building on the work he did when tapped by President Joe Biden in 2023 to be the informal envoy to the AI world, Obama has said he wants what he calls “a serious regulatory framework” in line with how the federal government sets standards for food or airlines. But Obama doesn’t have a specific plan that he’s pushing the 2028 contenders to sign onto.

Advisers to several prospective presidential candidates told CNN they see the polls showing the turn against data centers and social media, with some now coming to believe from their own conversations that more has already gone wrong than has been publicly disclosed — such as AI agents surreptitiously implanted across the internet. Yet they also admit they’re worried about being targeted by money if they speak out too much.

Outside groups linked to the AI industry have spent nearly $50 million on warring ads for 2026 races, according to data from AdImpact. And FEC records show leading AI PACs and nonprofits amassing significantly more, with groups linked to Anthropic and OpenAI combining to raise more than $150 million so far this cycle.

Alex Bores , a former Palantir employee who supported several state measures to regulate AI while serving in the New York assembly, was the subject of so many millions spent for and against him by AI-affiliated PACs this spring that the primary for an open congressional seat in Manhattan became one of the most expensive in history.

He said the prospect of campaign spending comes up regularly in meetings he’s already had with the candidates or teams of former Vice President Kamala Harris, former Transportation Secretary Pete Buttigieg, Gov. Josh Shapiro, Gov. JB Pritzker, Sen. Mark Kelly, Sen. Ruben Gallego, Sen. Jon Ossoff, Rep. Alexandria Ocasio-Cortez and Rep. Ro Khanna.

Michael M. Santiago/Getty Images

Bores, who started a group meant to push the 2028 candidates to take more definitive regulatory positions , told CNN the politicians and advisers keep bringing up the cash.

“‘What was it like having all that money spent against you?’ ‘How did you know they were going to come after you?’” Bores told CNN of the ways they’re “subtly and not so subtly asking, ‘Are there specific trip wires I can avoid?’ ‘How can I steer through these waters in a way that says the right thing about AI and not bring on the incredible amount of money from the other side?’”

“Some of the candidates who have longer odds are more willing to take risks,” Bores added. “Generally, a lot of people are still trying to find a way to appease everyone.”

But there’s money in both directions. Bores, for example, had $8.5 million spent against him by AI-linked PACs but another $22.5 million spent to boost him. One AI industry source, though, warned of Democrats chasing overly regulatory, doom-minded positions that could prompt a backlash against whomever emerges as the party nominee goes up against a Republican, much like the crypto millions that went for Trump in the 2024 election.

Jesse Hunt, a spokesperson for the Leading the Future PAC — largely funded by people with past or current ties to OpenAI and Andreessen Horowitz — said their focus remains innovation.

“We’ll continue to support leaders who want to find consensus on this important topic and reject extremism,” Hunt told CNN.

Also on the call list for many of the potential Democratic contenders trying to stake out AI positions: Khan, who since leaving the Biden administration has become the new nexus for corporate control thinking on the left and even some on the populist right.

Politicians talking about PACs or “p(doom)” — the shorthand for estimating the probability that AI will end humanity — are missing the point, Khan told CNN. Americans are sour on AI because they see their water and electricity rates going up due to the drain of data centers, or just from having a chat bot complicate their customer service experiences, let alone deny their health insurance claims.

In her conversations, Khan said she’s been urging prospective presidential candidates to remember that the AI questions are hitting against a backdrop of Americans feeling more economically precarious than ever while the rich keep getting richer.

“There’s also this basic question of, day after day, these AI executives are explaining how their technologies are hacking into other companies, hacking into other countries’ governments. That would be considered illegal if done by any random person,” she said. “And now that these companies and their very sophisticated technologies are doing it, why isn’t that illegal?”

Even for what little Democrats have put out, “it’s difficult for any of it to seem credible because Congress has been totally gutted of capacity, the executive branch has devolved into total corruption, the courts have become so libertarian,” Khan said. “And so there’s the way in which reckoning with these serious questions around AI and how do we regulate it is going to also require reckoning with the governing crisis.”

The generally more business-minded Raimondo, meanwhile, has through her new group, RAISE US, been pushing a more integrated approach to AI, focused on helping the workforce transition to an artificial intelligence economy.

Raimondo said she’s also been speaking with many of the prospective presidential candidates and has not been overwhelmed with any of what she has heard so far, classifying most as purely reactive to the pushback on data centers or the tweeted warnings of AI executives.

“If you want to be president, you have to step back and say, ‘You know, what’s America going to look like in a few years because of AI? And what is my plan proactively to make sure that we have the innovation and that every American worker has a good job in an AI world and everybody benefits from it?’” Raimondo told CNN. “When people are afraid, the worst thing to do is try to convince him that there’s nothing to worry about — and the best thing to do is to be honest, straight talk, accountable, and reassure them that you got it.”

Asked if she felt confident that the political dynamics will lead to the thoughtfulness and detail she believes AI demands, Raimondo said, “I’m a person of faith,” adding, “presidential candidates right, left, center are going to come to the realization that this could be the biggest issue and it won’t be enough to just be reactive.”

Raimondo had been toying with a presidential run in 2028 herself. She said that’s off the table as she digs in on this.

Pritzker and California Gov. Gavin Newsom have both spoken with multiple AI industry leaders and, along with Shapiro and Maryland Gov. Wes Moore, issued a new round of state executive orders in September to grapple with the technology.

The prospective presidential candidate who’s engaged the most on AI policy on the federal level, though, is Kelly. He released what he called a roadmap for AI a year ago built around a proposed federal fund, paid into by AI companies, to help pay for higher electricity bills, investing in infrastructure and safety protocols.

Walking to the Senate floor last week where he introduced a new bill to mitigate coming workforce disruptions, Kelly told CNN that he had been part of a briefing by an expert who came to see senators in September warning them how exponentially the technology is changing. Every conversation he has as part of the Armed Services and Intelligence committees, Kelly added, is packed with information about not just what’s happening in the US, but trying to keep up with adversaries, “and boy, every few months it’s something new.”

The Midterms’ Unlikely Villain: Data Centers | CNN 5 Things | CNN Audio

“I don’t think we’re approaching some kind of robot apocalypse here, you know, ‘Terminator’ stuff,” Kelly said, “but I think we are smart enough to be able to put the guardrails in place, and guide this in a direction that ultimately benefits humanity — that’s what I’d like to see.”

Kelly said among the people he’s spoken with in thinking through his ideas are Raimondo, former Bureau of Labor Statistics commissioner Erika McEntarfer, former acting Labor Secretary and current New York City Deputy Mayor Julie Su and American Federation of Teachers president Randi Weingarten. Obama’s been in his ear too, though Kelly said, “I don’t actually think we necessarily need that, especially if you pay attention to what’s going on in the world.”

Kelly added that he knows the threat of spending against him looms as he considers a 2028 run, but “we shouldn’t be making decisions based on that we think someone’s going to drop a money bomb on us for an election.”

To other Democratic leaders, there’s a political opportunity from explicitly going against the money.

“We have a decision to make: Do we want to stay quiet on AI and crypto and hope that they won’t spend money against us? Or do we want make their money an issue in the election?” said Connecticut Sen. Chris Murphy, who’s also discussed as a possible presidential candidate. “I think it’d be a huge mistake if we tried to curry favor with two industries that are not politically popular and are trying to get away with murder. So my view is that if we are very loud about the AI industry and the crypto industry’s attempts to try to buy our politics, that’s the safest political play we can make.”

CNN’s David Wright contributed to this report.

吉姆·冯德鲁斯卡/盖蒂图片社/文件

是的，那些可能参加 2028 年民主党总统竞选并与巴拉克·奥巴马交谈过的候选人一直表示，当他问及人工智能问题时，他们正在关注并制定计划。

奥巴马认为，就目前他所看到的而言，无论从问题的规模还是从政治角度来看，都远远不够。他把这一点告诉了他们。

这位前总统并非孤军奋战：从前联邦贸易委员会主席莉娜·汗到前商务部长吉娜·雷蒙多，各派民主党领导人都要求正在组建的联盟承担更多责任，他们认为迄今为止，联盟整体上更多的是被动应对，而非积极把握形势。但包括奥巴马在内的所有人，都未能就具体方案达成一致，一些2028年总统候选人也担心会触怒人工智能行业庞大的竞选资金。

尽管已离开白宫十年，奥巴马依然是民主党的标杆人物。他与新一代民主党领导人保持着密切联系，无论是他们主动寻求建议，还是他致电祝贺他们取得胜利，他都一视同仁。但据知情人士透露，奥巴马在加州之行、华盛顿办公室以及通过电话，也与各大人工智能公司的领导人——包括OpenAI的萨姆·奥特曼，以及谷歌和Anthropic的高管——以及众多外部专家、学者和非正式顾问进行了交流，其中包括曾任推特高管、后担任其数字战略办公室主任的杰森·戈德曼。

他酷爱阅读，尤其喜欢埃兹拉·克莱因、德里克·汤普森、查理·沃泽尔和凯西·牛顿的文章，并充当人工智能创始人与下一代政治新秀之间的非正式桥梁。这些创始人会向他汇报最新情况，并向他请教如何在华盛顿游刃有余，而新一代的政治新秀则会在迷茫不知所措时向他​​寻求建议。

周二，唐纳德·特朗普总统在白宫会见了多位人工智能领域的领军人物，并表示他们签署了一项新的承诺，旨在更好地“自我监管”各自公司的人工智能模型。（特朗普还表示，他将很快签署一项行政命令，将“人工智能”更名为“超级智能”。）华盛顿的大多数共和党人都效仿他的做法，反对联邦政府对人工智能进行广泛的监管。

朱莉娅·德马雷·尼金森/美联社

唐纳德·特朗普如此强烈反对人工智能安全防护措施的真正原因

与此同时，着眼于2028年大选的民主党人一直在发布演讲稿、成立委员会并制定计划——多位党内人士承认——这些都只是他们试图表明自己正在采取行动的初步尝试。但他们大多仍然不了解目前的技术发展水平，也不清楚两年内技术可能发生多大的变化。

几位潜在候选人的顾问私下向CNN承认，他们目前都在尽量避免任何可能被贴上标签的举动——他们进退两难，既不想显得过于受制于企业，又不想显得过于好斗；既不想显得不够民粹，又不想显得过于监管。他们也意识到，公开表态可能会让这些企业资金雄厚的超级政治行动委员会在竞争激烈的党内初选中将他们作为攻击目标。

奥巴马并未谈及人类末日，而是更关注失业、国家安全以及虚假信息泛滥对儿童（尤其是儿童）造成的影响。据多位知情人士透露，早在人工智能高管发出近期一系列警告之前，奥巴马就已多次向前来拜访或哪怕只是致电祝贺的民主党人强调，无论出于政策还是政治原因，他们都必须明确应对之策。

“鉴于我们可能面临的巨大变革，奥巴马总统认为，就这项技术的开发和应用展开务实且知情的对话至关重要，”他的高级顾问埃里克·舒尔茨告诉CNN。“它对我们的经济和民主的影响将取决于我们所做的选择。这意味着我们要了解人工智能带来的巨大机遇，同时也要清醒地认识到风险以及如何最好地降低这些风险。”

十多年来，奥巴马对人工智能的态度发生了转变：从最初在一次采访中开玩笑说“当你看到它即将发生时，你必须把墙上的电源拔掉”——到如今对美国如何毫无准备地应对即将到来的后果感到悲观。

本月初，他在科尔盖特大学的一次演讲中表示：“就我们如何治愈癌症或获得更好的能源而言，你无需拥有自主人工智能，也无需让它在互联网上自由漫游。之所以要这样做，是因为你必须推销一种人们愿意付费的产品。”

Kent Nishimura/AFP/Getty Images

奥巴马在多次接受CNN采访时表示，即便他认为美国政府能力出众，人工智能的崛起及其未来影响仍然会非常棘手。他补充说，他对特朗普缺乏信心——特朗普上周在联合国宣布他更愿意称之为“超级智能”，并在周二亲自会见了几位人工智能创始人，坚持其放任自流的态度——这使得应对人工智能的挑战更加艰巨。

奥巴马早在白宫时期就对技术变革表现出浓厚的兴趣，并在2023年被乔·拜登总统任命为人工智能领域的非正式特使后，进一步推动了相关工作。他曾表示，希望建立他所谓的“严格的监管框架”，使其与联邦政府制定食品或航空标准的方式相一致。但奥巴马并没有提出具体的计划，要求2028年总统候选人签署。

几位总统候选人的顾问告诉CNN，他们看到民调显示公众对数据中心和社交媒体的态度正在转变，一些人从与他人的交流中了解到，可能存在比公开披露的更多问题——例如人工智能程序被秘密植入互联网。但他们也承认，如果公开谈论太多，他们担心会成为金钱操纵的目标。

据AdImpact的数据显示，与人工智能行业相关的外部团体已在2026年总统大选的竞选广告上花费了近5000万美元。联邦选举委员会（FEC）的记录显示，主要的人工智能政治行动委员会（PAC）和非营利组织筹集的资金远不止于此，与Anthropologie和OpenAI相关的团体在本轮选举周期中迄今已筹集超过1.5亿美元。

亚历克斯·博雷斯曾是 Palantir 的一名员工，在纽约州议会任职期间，他支持了多项监管人工智能的州级措施。今年春天，与人工智能相关的政治行动委员会花费了数百万美元支持和反对他，导致曼哈顿一个空缺的国会席位的初选成为历史上最昂贵的初选之一。

他说，在他与前副总统卡马拉·哈里斯、前交通部长皮特·布蒂吉格、州长乔什·夏皮罗、州长JB·普里茨克、参议员马克·凯利、参议员鲁本·加列戈、参议员乔恩·奥索夫、众议员亚历山大·奥卡西奥-科尔特斯和众议员罗·卡纳的候选人或团队的会面中，竞选支出的前景经常被提及。

Michael M. Santiago/Getty Images

Bores 成立了一个旨在推动 2028 年候选人采取更明确的监管立场的组织，他告诉 CNN，政客和顾问们一直在谈论资金问题。

“‘被花那么多钱对付是什么感觉？’‘你怎么知道他们会来找你麻烦？’”博雷斯告诉CNN，他们“或明或暗地问：‘有没有什么特定的陷阱我可以避开？’‘我该如何妥善处理这些问题，既能正确看待人工智能，又不会招致对方巨额资金的攻击？’”

博雷斯补充说：“一些胜算较小的候选人更愿意冒险。总的来说，很多人仍在试图找到一种能让所有人都满意的方法。”

但资金的流向是双向的。例如，博雷斯曾被与人工智能相关的政治行动委员会花费850万美元反对，但也有2250万美元用于支持他。不过，一位人工智能行业人士警告说，民主党人过于追求监管和悲观的立场，可能会引发民众对最终胜出的党内候选人的强烈反对，就像2024年特朗普大选中遭遇的加密货币危机一样。

“引领未来”政治行动委员会的发言人杰西·亨特表示，他们的重点仍然是创新。该委员会主要由与 OpenAI 和 Andreessen Horowitz 有过去或现在联系的人资助。

亨特告诉 CNN：“我们将继续支持那些希望在这个重要议题上达成共识并反对极端主义的领导人。”

在许多试图在人工智能领域占据一席之地的潜在民主党候选人的通话名单上，还有一位是汗，自从离开拜登政府以来，他已成为左翼乃至一些民粹主义右翼企业控制思想的新中心人物。

汗告诉CNN，那些谈论政治行动委员会（PAC）或“末日概率”（p(doom)，用来估算人工智能将毁灭人类的概率的缩写）的政客们，完全没有抓住重点。美国人对人工智能感到反感，是因为他们看到数据中心消耗能源导致水费和电费上涨，或者仅仅是聊天机器人就让他们的客户服务体验变得复杂，更不用说还会拒绝他们的医疗保险索赔了。

在她的谈话中，汗表示，她一直在敦促有志成为总统的候选人记住，人工智能问题是在美国人感到经济状况比以往任何时候都更加不稳定，而富人却越来越富的背景下提出的。

“还有一个基本问题是，这些人工智能高管每天都在解释他们的技术是如何入侵其他公司、入侵其他国家政府的。如果换作普通人这样做，那肯定是非法的，”她说道。“而现在，是这些公司用他们非常先进的技术来做这件事，为什么就不违法了呢？”

汗表示，即便民主党人提出的方案寥寥无几，“也很难让人信服，因为国会的能力已被彻底削弱，行政部门已彻底腐败，法院也变得过于自由放任。” 他还说，“因此，要认真对待人工智能及其监管等这些严肃问题，就必须正视当前的治理危机。”

与此同时，更具商业头脑的雷蒙多通过她的新组织 RAISE US，一直在推动人工智能的更综合方法，重点是帮助劳动力向人工智能经济转型。

雷蒙多表示，她也与许多潜在的总统候选人进行了交谈，但到目前为止，她听到的任何内容都没有让她感到震惊，她认为大多数候选人的言论纯粹是对数据中心遭到抵制或人工智能高管在推特上发出的警告的回应。

“如果你想当总统，你必须退后一步，扪心自问：‘你知道吗，几年后人工智能会给美国带来什么？我有什么积极主动的计划，确保我们拥有创新能力，确保每个美国工人都能在人工智能时代找到好工作，让每个人都能从中受益？’”雷蒙多告诉CNN。“当人们感到恐惧时，最糟糕的做法是试图说服他们没什么好担心的——而最好的做法是坦诚直率、勇于担当，并让他们相信你能掌控局面。”

当被问及她是否相信政治动态会带来她认为人工智能所需要的深思熟虑和细致入微时，雷蒙多说：“我是一个有信仰的人，”并补充道，“总统候选人，无论左翼还是中间派，都会意识到这可能是最大的问题，仅仅被动应对是不够的。”

雷蒙多本人曾考虑过在2028年竞选总统。但她表示，随着她全力投入到当前事务中，竞选总统的计划已经搁置。

普里茨克和加州州长加文·纽森都与多位人工智能行业领袖进行了交谈，并与夏皮罗和马里兰州州长韦斯·摩尔一起，在9月份发布了一系列新的州行政命令，以应对这项技术。

不过，在联邦层面，最积极参与人工智能政策的总统候选人当属凯利。一年前，他发布了一份他称之为“人工智能路线图”的方案，其核心内容是提议设立一项联邦基金，由人工智能公司出资，用于支付不断上涨的电费，并投资于基础设施和安全协议。

上周，凯利在参议院提出一项旨在缓解即将到来的劳动力市场动荡的新法案时告诉CNN，他曾参加过一位专家9月份向参议员们汇报的简报会，这位专家警告他们技术变革正以惊人的速度发展。凯利补充说，他在军事委员会和情报委员会的每一次谈话都充满了信息，不仅涉及美国正在发生的事情，还涉及如何跟上对手的步伐，“哎，每隔几个月就会出现新的情况。”

中期选举中意想不到的“反派”：数据中心 | CNN 五件事 | CNN音频

“我不认为我们正在走向某种机器人末日，你知道，就像《终结者》里那样，”凯利说，“但我认为我们足够聪明，能够设置护栏，引导它朝着最终造福人类的方向发展——这正是我希望看到的。”

凯利表示，为了完善自己的想法，他曾与雷蒙多、前劳工统计局局长埃里卡·麦克恩塔弗、前代理劳工部长兼现任纽约市副市长朱莉·苏以及美国教师联合会主席兰迪·温加滕等人进行过交流。奥巴马也曾向他提出建议，不过凯利说：“我其实并不认为我们真的需要奥巴马的建议，尤其是在你关注世界局势的情况下。”

凯利补充说，他知道在考虑参加 2028 年总统竞选时，有人会用金钱攻击他，但他“不应该基于这种想法来做决定，因为我们认为有人会在选举中向我们投入巨资”。

对其他民主党领导人来说，公开反对金钱势力存在着政治机会。

“我们面临一个抉择：是保持对人工智能和加密货币的沉默，寄希望于它们不会花钱攻击我们？还是让它们花钱的问题成为选举中的一个焦点？”康涅狄格州参议员克里斯·墨菲说道，他也是总统候选人的热门人选之一。“如果我们试图讨好这两个在政治上不受欢迎、却又想逍遥法外的行业，那将是一个巨大的错误。所以我认为，如果我们大声疾呼人工智能和加密货币行业试图收买我们政治的行为，这才是最稳妥的政治策略。”

CNN的戴维·赖特对本报道亦有贡献。