Congress has failed to regulate AI for years – and there’s little sign that’s changing

In December 2024, a bipartisan congressional task force studying the possible perils of artificial intelligence issued 85 recommendations for how lawmakers could act on the issue. GOP Rep. Jay Obernolte, a key figure on the task force, asked Republican leaders to go even further, establishing a special committee to draft legislation.

But almost two years later, there are no real laws to rein in the rapid advancement of AI, and the Trump administration has made clear it has no appetite for regulation. Obernolte told CNN he couldn’t even get his special committee approved by GOP leaders, who a source said felt the task force had accomplished what it set out to do and that existing committees could work from its recommendations.

“I think the need for action is urgent,” Obernolte said in an interview. “We have been saying this for the last year and a half.”

The episode is emblematic of how AI has vexed US lawmakers, the complicated nature of the issue colliding head on with the intense partisanship and convoluted bureaucracy of Congress. Multiple bipartisan bills have been introduced , including a proposal to require the most powerful AI companies develop a kill switch for their technology and legislation to establish safeguards for all AI models.

Many proposals have never even made it to the floor for a vote, waylaid in a morass of committees. Lawmakers openly talk about how some of their older colleagues don’t understand the technology – and express open skepticism about their ability to effectively legislate it.

“If people put all of their hopes in government and Congress to solve the problems in the world, I think that they will be woefully disappointed,” GOP Rep. Eric Burlison, who served on the now-disbanded AI task force, told CNN.

The issue of AI has once again risen to the center of public debate in recent weeks , with whistleblower and AI executives raising alarm bills and pleading for regulation, while President Donald Trump , House Speaker Mike Johnson and others warn that slowing down development of the technology will only serve to benefit China in what effectively amounts to an arms race.

Earlier this week, members of the House Committee on Science, Space and Technology met with representatives of Hugging Face, Anthropic, Open AI and other third party auditors to brief lawmakers, leaders of the committee said. The representatives were asked if they need oversight or if they can self-police, and they said they were committed to instituting guardrails but they couldn’t trust that everyone in the industry would do the same, according to two sources in the meeting.

Senate Majority Leader John Thune has called for “a light touch approach.” Democrats generally prefer a more aggressive posture.

“I am worried about the most powerful technology in the history of our world, of humanity, being on a collision course with congressional dysfunction,” Democratic Rep. Lori Trahan, who has been working on AI legislation with Obernolte for over a year, told CNN.

Chip Somodevilla/Getty Images/File

Since the 2024 task force report, Congress has taken only steps around the margins of AI. A senior GOP leadership aide said they did not proceed with a special committee because the broad view among leadership was that the task force had accomplished what it set out to do and the standing committees that already have AI jurisdiction had the recommendations from the task force to consider.

Last year, Trump signed into law a bill that made it illegal to share online nonconsensual, explicit images — real or computer-generated — that also required tech platforms to remove such images within 48 hours of being notified about them. On Wednesday, the House overwhelmingly passed a bill to limit the rising cost of utility bills linked to the rapid expansion of data centers that power AI.

But more substantive bills have gone nowhere. The House Committee on Science, Space and Technology passed 10 AI-related bills out of committee in June that still have not been brought to the floor. Lawmakers in both parties – including Thune, Democratic Rep. Amy Klobuchar, GOP Sen. Josh Hawley and Democratic Sen. Richard Blumenthal – have put forward proposals. Sen. Bernie Sanders, who caucuses with Democrats, has stridently warned in recent days about the dangers of AI. But there is no agreed-upon leader or expert on the issue in Capitol Hill.

“Tell me what congressman is going to educate every other congressman on what Anthropic does. Give me a break. Half the Congress is 70 years old. They don’t even know how to turn on their cell phone without help from staff,” GOP Rep. Rich McCormick of Georgia, who served on the task force with Obernolte, told CNN.

Trump has long touted the benefits of AI, but other Republicans had been wary of over-regulation before he took office. House Majority Leader Steve Scalise told members of the task force in 2024 that he was not interested in seeing heavy regulations stifling AI, according to a source familiar with the discussion. Scalise said publicly at the time that he wanted to fill in gaps but not add burdensome regulations.

Chip Somodevilla/Getty Images.File

Some who favor regulation see their cause buoyed by warnings from AI industry leaders – even if they’re wary anything will happen.

“My message is like, ‘Hey, thank goodness. Welcome to the party.’ I’m glad everybody’s getting on board that this now has the national attention that it deserves,’” GOP Rep. Scott Franklin, a former member of the task force and co-sponsor on a number of pieces of AI legislation, told CNN. “But this is something we’ve been working on legislatively for a long time trying to get the attention to have these issues looked at.”

GOP Rep. Kat Cammack was even more pointed.

“We certainly have the capability to do it. I don’t know if the political will is there,” she told CNN. “I think the political will doesn’t exist because you have members that don’t really understand it.”

But others say their skepticism remains – including of the unusual calls from the AI industry for congressional intervention.

“I’m not bailing you out,” McCormick told CNN. “Why am I responsible for you being safe? That’s good business. Being safe is good business. You put the guardrails on yourself.”

Democrats, meanwhile, have intensified their calls for regulation in recent weeks, particularly after former President Barack Obama called on his party to prioritize the issue . After the task force concluded its work in 2024, the top Democrat in the House, Rep. Hakeem Jeffries, created his own commission. The party has generally sought to frame the issue as one of the everyman vs. the elites.

“I think that for far too long, big tech and money in politics have driven agendas in Congress that have crippled our ability to meet the moment on every urgent issue facing the country. So, we can’t rely on AI companies to lead the way on this, to regulate themselves. This is the role of the federal government. And it is absolutely essential that we rise to meet this moment,” Democratic Rep. Brittany Pettersen said.

Democrats, too, suggest the issue of regulation is not as complicated as some of their GOP colleagues suggest. Competitiveness with China is an issue, but what good is competition in a world run by AI? Sure, some lawmakers are aging – but lawmakers have never had to be subject matter experts for every bill they support.

“If you took that standard, then Congress could never regulate airplanes or medicine or energy,” said Democratic Rep. Ted Lieu, who was the top Democrat on the bipartisan task force and has continued the work in the minority even without a GOP counterpart. “There’s all sorts of regulations that are common sense that Congress can pass and should pass and doesn’t require anyone being a [computer science] major to understand what these bills do.”

A bipartisan group of members have called on Johnson to keep the House in session so lawmakers can work on negotiating AI reforms. That’s not happening: Johnson decided instead to send the House home a day early, not to return until after the November midterms.

But some say it’s not reason to give up. Democratic Rep. Don Beyer, one of the lawmakers calling on Johnson to keep the House in session, found himself consoling two of his constituents earlier this week who have a fourth-grade daughter and are worried there will not be a future for her if the threats of AI come true.

“I said to them: I think there is still time.”

2024年12月，一个由两党议员组成的国会工作组就人工智能可能带来的风险提出了85项建议，旨在指导立法者如何应对这一问题。共和党众议员杰伊·奥伯诺尔特是该工作组的关键成员，他要求共和党领导人采取更进一步的行动，成立一个特别委员会来起草相关立法。

但近两年过去了，仍然没有真正有效的法律来遏制人工智能的快速发展，而特朗普政府也明确表示无意进行监管。奥伯诺尔特告诉CNN，他甚至无法让共和党领导人批准他成立的特别委员会。一位消息人士称，共和党领导人认为该工作组已经完成了既定目标，现有委员会可以根据其建议开展工作。

奥伯诺尔特在接受采访时表示：“我认为现在迫切需要采取行动。过去一年半以来，我们一直在强调这一点。”

这一事件充分体现了人工智能如何令美国立法者感到棘手：问题的复杂性与国会激烈的党派之争和繁琐的官僚程序正面交锋。目前已提出多项两党法案，其中包括一项要求实力最强的人工智能公司为其技术开发“终止开关”的提案，以及一项旨在为所有人工智能模型建立保障措施的立法。

许多提案甚至从未提交议会表决，就被困在了错综复杂的委员会中。议员们公开谈论一些资深同事不了解这项技术，并对他们能否有效地将其立法表示怀疑。

“如果人们把所有希望都寄托在政府和国会身上，指望他们解决世界上的问题，我认为他们会非常失望，”曾任现已解散的人工智能工作组成员的共和党众议员埃里克·伯利森告诉 CNN。

近几周来，人工智能问题再次成为公众辩论的焦点，举报人和人工智能高管发出警告，呼吁进行监管，而唐纳德·特朗普总统、众议院议长迈克·约翰逊等人则警告说，放慢这项技术的发展速度只会让中国在实际上相当于军备竞赛的局面中受益。

本周早些时候，众议院科学、空间和技术委员会的成员会见了Hugging Face、Anthropic、OpenAI和其他第三方审计机构的代表，向议员们汇报了情况。委员会领导人表示，议员们被问及是否需要监管，或者是否可以进行自我监管。据两位与会人士透露，这些代表表示，他们致力于建立监管机制，但无法保证业内所有机构都会这样做。

参议院多数党领袖约翰·图恩呼吁采取“温和的方式”。而民主党人通常更倾向于采取强硬姿态。

“我担心我们世界、人类历史上最强大的技术，会与国会的失能发生冲突，”民主党众议员洛里·特拉汉 (Lori Trahan) 告诉 CNN，她与奥伯诺尔特 (Obernolte) 合作人工智能立法已有一年多的时间。

Chip Somodevilla/Getty Images/文件

自2024年工作组报告发布以来，国会在人工智能领域仅采取了一些无关紧要的措施。一位共和党高级领导层助手表示，他们没有成立特别委员会，因为领导层普遍认为工作组已经完成了既定目标，而且已有负责人工智能事务的常设委员会也已收到工作组的建议，可以进行审议。

去年，特朗普签署了一项法案，该法案禁止在网上分享未经同意的露骨图片——无论是真实的还是电脑生成的——并要求科技平台在接到通知后48小时内删除此类图片。周三，众议院以压倒性多数通过了一项法案，旨在限制因人工智能数据中心快速扩张而导致的公用事业费用上涨。

但更具实质性的法案却毫无进展。众议院科学、空间和技术委员会6月份通过了10项与人工智能相关的法案，但至今仍未提交全体会议审议。包括图恩、民主党众议员艾米·克洛布查尔、共和党参议员乔什·霍利和民主党参议员理查德·布卢门撒尔在内的两党议员都提出了相关提案。与民主党人结盟的参议员伯尼·桑德斯近日也对人工智能的危险性发出了强烈的警告。然而，国会山目前尚未就此问题达成共识，既没有一位权威的领导者，也没有一位专家。

“告诉我，哪位国会议员会去向其他国会议员普及安人智公司（Anthropic）的业务？别开玩笑了。国​​会一半的议员都70岁了。他们甚至连怎么打开手机都得靠工作人员帮忙。”曾与奥伯诺尔特一同在工作组任职的佐治亚州共和党众议员里奇·麦考密克（Rich McCormick）告诉CNN。

特朗普长期以来一直大力宣传人工智能的益处，但在他上任之前，其他共和党人就对过度监管持谨慎态度。据一位知情人士透露，众议院多数党领袖史蒂夫·斯卡利斯在2024年告诉工作组成员，他不希望看到繁重的监管扼杀人工智能的发展。斯卡利斯当时公开表示，他希望填补监管空白，而不是增加繁琐的监管措施。

Chip Somodevilla/Getty Images.File

一些支持监管的人士认为，人工智能行业领导者的警告鼓舞了他们的事业——即使他们也怀疑最终是否会有任何改变。

“我想说的是，‘嘿，谢天谢地，欢迎加入这场讨论。’我很高兴大家都能关注到这个问题，它现在终于得到了应有的全国性关注，”共和党众议员斯科特·富兰克林告诉CNN。富兰克林曾是该工作组成员，也是多项人工智能法案的共同发起人。“但我们长期以来一直在立法层面努力，试图引起人们对这些问题的重视。”

共和党众议员凯特·卡马克 (Kat Cammack) 的言辞则更加尖锐。

“我们当然有能力做到这一点。我不知道是否有政治意愿，”她告诉CNN。“我认为缺乏政治意愿是因为有些议员并不真正理解这一点。”

但也有人表示他们仍然持怀疑态度——包括对人工智能行业要求国会干预的罕见呼吁。

“我不会出手相救，”麦考密克告诉CNN。“我为什么要负责你的安全？这是生意之道。安全才是生意之道。你自己给自己设限。”

与此同时，民主党近几周加大了对监管的呼吁力度，尤其是在前总统奥巴马呼吁民主党将此问题列为优先事项之后。在特别工作组于2024年结束工作后，众议院民主党领袖哈基姆·杰弗里斯议员成立了自己的委员会。该党通常试图将此问题定义为普通民众与精英阶层之间的冲突。

“我认为，长期以来，大型科技公司和政治资金一直在左右国会的议程，严重削弱了我们应对国家面临的每一个紧迫问题的能力。因此，我们不能指望人工智能公司在这方面发挥领导作用，进行自我监管。这应该是联邦政府的职责。我们必须挺身而出，迎接这一挑战。”民主党众议员布列塔尼·彼得森说道。

民主党人也认为，监管问题并不像一些共和党同僚所说的那么复杂。与中国的竞争固然重要，但在一个人工智能主导的世界里，竞争又有何意义呢？诚然，一些议员年纪大了——但他们支持的每一项法案，议员们都不必是相关领域的专家。

“如果按照这个标准，国会就永远无法监管飞机、药品或能源了，”民主党众议员特德·刘说道。他是两党工作组的首席民主党成员，即使没有共和党同僚，他也继续以少数党议员的身份推进这项工作。“有很多符合常识的监管措施，国会可以而且应该通过，这些措施并不需要计算机科学专业出身的人才能理解。”

一个由两党议员组成的小组呼吁约翰逊让众议院继续开会，以便议员们能够就人工智能改革进行谈判。但这一呼吁并未实现：约翰逊决定提前一天让众议院休会，直到11月中期选举之后才会复会。

但有些人认为这并非放弃的理由。民主党众议员唐·拜尔是呼吁约翰逊让众议院继续开会的议员之一。本周早些时候，他不得不安慰两位选民，这两位选民的女儿正在读四年级，他们担心如果人工智能的威胁成真，女儿将没有未来。

我对他们说：我觉得还有时间。