AI PACs Pour $300M Into US Midterm Primaries to Dictate Federal Rules

Bloomberg Originals

Tech firms are pouring hundreds of millions of dollars into congressional primaries as insurance to preempt local regulation and protect a trillion dollars in infrastructure, even as grassroots voter backlash over utility costs and data centers mounts.

AI-assisted summary of Why Big AI Is Pouring Millions Into the Midterms — original source. Check the original for full context.

Source published: 2026-10-16

Key ideas

AI-aligned political action committees have raised nearly $300 million to sway congressional races, contributing to a record-breaking $11 billion in total political advertising across the midterms. (0:01)
The AI lobby is split between the deregulatory super PAC Leading the Future ($140 million) and Anthropic-backed Public First Action ($83 million), which advocates for strict safety guardrails. (2:23)
With tech firms committing an estimated $1 trillion toward data centers and core infrastructure, multi-million-dollar super PAC investments function as a modest insurance policy to secure favorable regulation. (2:23)
Grassroots resistance to data centers is escalating over utility costs and land use, highlighted by a Virginia appellate ruling that blocked a planned 35-center project spanning nearly 2,000 acres. (7:27)

Detailed notes

## Background Published by Bloomberg Originals on YouTube on October 16, 2026, this report examines how the tech sector is spending hundreds of millions of dollars to influence the US midterm elections. According to the description, industry groups are pouring money into congressional campaigns to secure favorable rule-making, even as voter anxiety intensifies over soaring energy costs, job losses, sprawling data centers, and existential risk.

## TL;DR Facing legislative inertia in Washington, the artificial intelligence sector has funneled nearly $300 million into congressional midterms to elect lawmakers who will author federal regulations. The industry is split between competing super PACs: deregulatory groups like Leading the Future ($140 million) pushing light-touch rules to outpace China, and safety-oriented factions like Anthropic-backed Public First Action ($83 million) demanding strict guardrails. Using aggressive persuasion tactics reminiscent of Big Tobacco and crypto's Fairshake PAC, these campaigns target party primaries to sway safe districts. However, this corporate political blitz is colliding with intense grassroots resistance against local data center construction as well as dire warnings from former industry researchers regarding human extinction.

## Atmosphere & style The production delivers a sleek, investigative tone that blends data-driven political reporting with an undercurrent of systemic dread. It moves between high-stakes financial breakdowns of PAC spending and grounded, contentious community battles against humming power substations in suburban backyards. The narrative feels urgent and sobering, juxtaposing calculated tech-industry lobbying strategies against stark whistleblower warnings to question whether democratic representation can withstand an unprecedented flood of corporate capital.

## Bulletin - [00:01](https://www.youtube.com/watch?v=wraPuftwsxQ&t=1s) AI-aligned political action committees have raised nearly $300 million to sway congressional races, contributing to a record-breaking $11 billion in total political advertising across the midterms. - [02:23](https://www.youtube.com/watch?v=wraPuftwsxQ&t=143s) The AI lobby is split between the deregulatory super PAC Leading the Future ($140 million) and Anthropic-backed Public First Action ($83 million), which advocates for strict safety guardrails. - [02:23](https://www.youtube.com/watch?v=wraPuftwsxQ&t=143s) With tech firms committing an estimated $1 trillion toward data centers and core infrastructure, multi-million-dollar super PAC investments function as a modest insurance policy to secure favorable regulation. - [07:27](https://www.youtube.com/watch?v=wraPuftwsxQ&t=447s) Grassroots resistance to data centers is escalating over utility costs and land use, highlighted by a Virginia appellate ruling that blocked a planned 35-center project spanning nearly 2,000 acres.

## Key takeaways - [00:01](https://www.youtube.com/watch?v=wraPuftwsxQ&t=1s) Artificial intelligence has emerged as a major cross-partisan election issue, driven by public anxiety over job displacement, energy consumption, and corporate overreach. - [00:01](https://www.youtube.com/watch?v=wraPuftwsxQ&t=1s) Federal legislative inertia and debates over technological competition with China have spurred record-breaking midterm ad spending exceeding $11 billion. - [00:01](https://www.youtube.com/watch?v=wraPuftwsxQ&t=1s) AI-aligned political action committees amassed nearly $300 million to target federal races and install lawmakers friendly to tech industry priorities. - [02:23](https://www.youtube.com/watch?v=wraPuftwsxQ&t=143s) Rather than relying solely on conventional lobbying, tech super PACs increasingly intervene in party primaries within safe districts to pre-select sympathetic legislators. - [02:23](https://www.youtube.com/watch?v=wraPuftwsxQ&t=143s) Leading the Future raised $140 million to advocate for a light-touch national framework designed to preempt fragmented state laws and outpace China. - [02:23](https://www.youtube.com/watch?v=wraPuftwsxQ&t=143s) Public First Action, funded with an estimated $83 million from Anthropic, operates as a rival super PAC backing candidates who favor strict AI safety guardrails. - [02:23](https://www.youtube.com/watch?v=wraPuftwsxQ&t=143s) Relative to the roughly $1 trillion invested annually in data center infrastructure, political donations represent a high-yield hedge to lock in beneficial rules of governance. - [04:30](https://www.youtube.com/watch?v=wraPuftwsxQ&t=270s) Campaign persuasion advertising is increasingly produced and micro-targeted using the same generative AI tools that industry groups seek to protect from regulation. - [04:30](https://www.youtube.com/watch?v=wraPuftwsxQ&t=270s) Observers liken the AI lobby's rhetorical approach to Big Tobacco's historical playbook of backing weak national standards to preempt stricter state-level enforcement. - [04:30](https://www.youtube.com/watch?v=wraPuftwsxQ&t=270s) AI groups are emulating the blueprint of the 2024 crypto super PAC Fairshake, which demonstrated how concentrated capital can defeat adversarial primary candidates. - [04:30](https://www.youtube.com/watch?v=wraPuftwsxQ&t=270s) Unlike cryptocurrency, AI generates acute physical externalities—including power grid strain, noise pollution, and water use—that make public backlash harder to blunt with advertising. - [07:27](https://www.youtube.com/watch?v=wraPuftwsxQ&t=447s) Suburban and rural communities are mounting legal challenges against data centers, securing major victories such as halting a 2,000-acre development near Manassas National Battlefield Park. - [07:27](https://www.youtube.com/watch?v=wraPuftwsxQ&t=447s) Dissent within frontier labs has intensified, with former researchers such as Jacob Coxon warning that unchecked capabilities pose catastrophic extinction risks. - [10:46](https://www.youtube.com/watch?v=wraPuftwsxQ&t=646s) The central question facing Washington is whether massive corporate expenditures will dictate emerging AI policy or whether lawmakers will prioritize constituent protections.

## Detailed walkthrough ### [00:01](https://www.youtube.com/watch?v=wraPuftwsxQ&t=1s) The Most Expensive Midterm and the Rise of AI PACs

The midterm election is colliding with an unprecedented technological revolution, as artificial intelligence emerges as an issue cutting across traditional partisan lines. Unlike standard campaign wedge topics, AI touches everything from the economy and the environment to the existential future of humanity. Voters across the country are expressing profound apprehension regarding the pace of deployment, particularly fears of widespread white-collar and blue-collar job displacement, erosion of personal autonomy, and the rapid construction of energy-hungry data centers in local communities. This tension reflects a broader populist undercurrent against tech elites who are seen as unilaterally imposing unproven systems on the public, sparking grassroots protests outside corporate headquarters like OpenAI.

Political leadership in Washington remains thoroughly fragmented. While some officials stress the imperative of winning the technological race against China—warning that whoever dominates AI will dictate global geopolitical power—others demand international cooperation, self-regulation, or sweeping guardrails to address regulatory vacuums. Because of this legislative inertia, corporate interests have transformed the midterms into the most expensive election cycle in American history, with political advertising spending surpassing $11 billion. Spearheading this historic influx of capital are newly established AI-aligned political action committees, which have amassed nearly $300 million to influence federal races and ensure sympathetic lawmakers control the regulatory levers.

“Poll after poll after poll shows that there's a lot of anxiety among the general population about AI and what it's going to mean for their futures.”

- Total political advertising across the midterms exceeded $11 billion, breaking all prior spending records. - Political groups aligned with the AI sector raised nearly $300 million to intervene in congressional races. - The dominant public concerns driving voter backlash include job losses, loss of human autonomy, elite technological imposition, and the local environmental footprint of infrastructure.

---

[02:23](https://www.youtube.com/watch?v=wraPuftwsxQ&t=143s) The Super PAC Clash: Leading the Future vs. Public First Action

Faced with a gridlocked Congress where passing substantive legislation is notoriously difficult, the tech sector has pivoted from conventional lobbying to the electoral process itself. The strategy focuses on intervening directly in party primaries within safe districts to install candidates predisposed to their policy preferences before the general election even begins. However, the AI lobby is fundamentally divided into two warring camps, both claiming to be pro-technology but offering diametrically opposed regulatory blueprints. One faction advocates for minimal, light-touch federal oversight designed to preempt state-level restrictions, while the other demands stringent safety standards and strict corporate guardrails.

```mermaid flowchart TD accTitle: Competing AI Super PAC Strategies accDescr: Diagram illustrating the two opposing AI super PAC factions, their funding sources, and their regulatory objectives.

subgraph LTF["Leading the Future PAC (~$140M)"] direction TB L1["Light-touch federal rules"] L2["Preempt state legislation"] L3["Outpace China's capabilities"] end

subgraph PFA["Public First Action (~$83M)"] direction TB P1["Backed by Anthropic"] P2["Led by Brad Carson"] P3["Strict AI safety guardrails"] end

LTF -->|"Primary intervention"| Cong["US Congress"] PFA -->|"Primary intervention"| Cong ```

Leading the Future emerged as the flagship vehicle for the deregulatory faction, raising $140 million to elect candidates who oppose heavy-handed guardrails and argue that fragmented state laws could paralyze American competitiveness against China. In response, a rival super PAC called Public First Action was formed to counter that deregulatory agenda. Spearheaded by Brad Carson and funded with an estimated $83 million from safety-focused AI lab Anthropic, Public First Action targets both Republican and Democratic primaries to elevate candidates who champion the public interest. Industry insiders view this massive political expenditure as modest relative to the stakes: with tech firms committing a trillion dollars toward data center construction and hundreds of billions into core development, an investment of roughly $100 million in electoral politics represents a nominal insurance policy to secure favorable rules of governance.

“The way that you really influence Congress is by intervening in primaries and electing candidates who will win safe districts who will be friendly to your agenda,” explains Brad Carson.

- **Leading the Future:** Raised approximately $140 million; lobbies for a light-touch national framework to block piecemeal state laws and prevent China from leading the AI race. - **Public First Action:** Backed by Anthropic with an estimated $83 million; supports candidates across both parties who favor strict safety and public-interest regulations. - **Capital Scale:** The tech industry is pouring around $1 trillion into data center infrastructure alone in the current year, making political spending a high-yield investment to protect capital.

---

[04:30](https://www.youtube.com/watch?v=wraPuftwsxQ&t=270s) Persuasion Spending, the Tobacco Playbook, and Fairshake's Precedent

The entirety of this political capital is funneled directly into persuasion spending—primarily media advertising—which is now being created and micro-targeted using the very generative AI systems the campaigns seek to protect. This dynamic makes deceptive or hyper-realistic political messaging faster and cheaper to manufacture than ever before. Analysts observe that the modern AI lobby's rhetorical tactics closely mimic the historic playbook pioneered by Big Tobacco: paying lip service to federal standards while actively fighting any enforcement mechanism, insisting on national preemption to nullify stricter state statutes, professing concern for harms while encouraging adoption, and aggressively sidestepping legal liability.

The structural blueprint for this aggressive intervention stems from the landmark *Citizens United* ruling, which eliminated limits on independent corporate expenditures and caused each successive election cycle to eclipse the last in total cost. In 2024, the cryptocurrency industry established the Fairshake super PAC, amassing an enormous war chest to systematically carpet-bomb candidates hostile to crypto and purge them from office. While the AI sector attempts to emulate Fairshake's scorched-earth model, AI faces a far more hostile public landscape. While crypto was an insular financial issue largely unified behind specific deregulatory perks, AI directly disrupts physical neighborhoods, consumes scarce municipal utilities, threatens white-collar livelihood, and defies neat partisan divides.

“You're not trying to persuade them of, of how good your industry is. You're trying to show them that you have power to spend money and influence the outcome of their election.”

- Persuasion advertising accounts for 100% of the super PAC expenditures. - The 2024 crypto PAC Fairshake proved that concentrated industry capital could systematically defeat unfriendly politicians in congressional primaries. - Unlike crypto, AI creates immediate physical externalities—grid strain, noise, land use, and widespread labor anxiety—that make public opinion far harder to sway through pure advertising.

---

[07:27](https://www.youtube.com/watch?v=wraPuftwsxQ&t=447s) Backyard Battles and Existential Rifts Over Data Centers

The political theoretical debate over AI regulation crashes into physical reality at the municipal level, where residents in suburban and rural communities are rebelling against sprawling data center developments. In neighborhoods originally promised retail parks and open green spaces, residents have instead found themselves living next door to massive industrial complexes and humming electrical substations. The centers are notorious for draining regional water supplies, driving up residential electricity bills, and bulldozing pristine woodlands. In Northern Virginia, community activist Roger Yackel and fellow residents organized against a plan to build more than 35 data centers sprawling across nearly 2,000 acres adjacent to Manassas National Battlefield Park, winning a unanimous victory before the Court of Appeals to void the development.

Local animosity illustrates the limits of PAC spending: multi-million-dollar advertising campaigns cannot easily erase voter fury over soaring utility rates or industrial encroachment. At the same time, the AI industry remains deeply fractured from within regarding the actual hazards of frontier models. While enterprise leaders publicly testify that their technologies are safe and promise to mitigate energy costs, senior safety researchers are abandoning the industry in protest. Jacob Coxon, a former researcher at both OpenAI and Anthropic, resigned over deep alarm regarding uncontrolled frontier capabilities. The schism has left lawmakers grappling with whether letting tech leaders draft their own legislation represents an obvious case of the fox guarding the henhouse.

“On the current path, I think it is more likely than not that humanity loses control to these AIs and it could end in human extinction,” warns former researcher Jacob Coxon.

- **Manassas Battle:** Citizen organizing led by Roger Yackel successfully overturned a massive project near Manassas National Battlefield Park that sought to place over 35 data centers across nearly 2,000 acres. - **Resource Strains:** Local pushback is primarily fueled by skyrocketing electricity rates, substantial water consumption for cooling systems, noise, and the visual blight of massive electrical substations. - **Internal Whistleblowers:** Elite safety researchers from top frontier labs continue to exit the private sector, warning that humanity is currently on track to lose control of advanced AI systems.

[10:46](https://www.youtube.com/watch?v=wraPuftwsxQ&t=646s) The Battle Over Who Writes AI Regulations

The central conflict surrounding artificial intelligence governance ultimately comes down to an unresolved power struggle between corporate interests and the public good. Reaching an equitable balance between these competing forces presents what is perhaps the hardest challenge of all: deciding whether future legal frameworks are designed primarily to benefit the AI industry or the people it impacts. At its core, the flood of political money pouring into the midterm elections highlights a fight for institutional control over emerging technology policy, raising fundamental questions about who will author the legal landscape for decades to come.

As the speaker emphasizes: “What's at stake is who's going to write the rules for the future, and that's really what a lot of this comes down to.” The sheer scale of corporate spending leaves open the question of whether tech dollars will be the decisive factor steering national AI regulation, or whether lawmakers will retain the political independence needed to listen to the people they represent. Framing the ultimate test facing Washington, the speaker asks: “This corporate spending, is that going to be what determines the future of AI regulation in this country or are politicians going to be able to stake out their own positions based on what their constituents want?”

- **The Core Dilemma:** Deciding whether emerging AI regulations will tilt in favor of corporate industry priorities or public protections. - **The Political Stake:** Determining whether unprecedented campaign spending will set the regulatory agenda or if elected representatives will champion voter concerns regarding energy consumption, job displacement, and safety risks.

## Notable quotes > [02:23](https://www.youtube.com/watch?v=wraPuftwsxQ&t=143s) The way that you really influence Congress is by intervening in primaries and electing candidates who will win safe districts who will be friendly to your agenda, — Brad Carson

> [07:27](https://www.youtube.com/watch?v=wraPuftwsxQ&t=447s) On the current path, I think it is more likely than not that humanity loses control to these AIs and it could end in human extinction, — Jacob Coxon

## Mentioned Leading the Future, Public First Action, Anthropic, OpenAI, Brad Carson, Jacob Coxon, Roger Yackel, Fairshake, Citizens United, Manassas National Battlefield Park, US Congress, China, Bloomberg Originals

## Video description AI is a major force in the 2026 US midterm elections, with industry groups spending hundreds of millions of dollars to elect candidates who might write the rules governing the controversial technology. But this unprecedented flood of money is splashing against a backdrop of growing public anxiety around AI’s rising energy costs, potential job destruction, sprawling data centers and existential risk.

-------- Like this video? Subscribe: http://www.youtube.com/Bloomberg?sub_confirmation=1

Get unlimited access to Bloomberg.com for just $1.99 your first month: https://www.bloomberg.com/subscriptions?in_source=YoutubeOriginals Bloomberg Originals offers bold takes for curious minds on today’s biggest topics. Hosted by experts covering stories you haven’t seen and viewpoints you haven’t heard, you’ll discover cinematic, data-led shows that investigate the intersection of business and culture. Exploring every angle of climate change, technology, finance, sports and beyond, Bloomberg Originals is business as you’ve never seen it.

Subscribe for business news, but not as you've known it: exclusive interviews, fascinating profiles, data-driven analysis, and the latest in tech innovation from around the world.

Visit our partner channel Bloomberg News for global news and insight in an instant.

[Source](https://www.youtube.com/watch?v=wraPuftwsxQ)

中文简报

AI超级政治行动委员会注资3亿美元干预美国中期选举初选，意图主导联邦监管规则科技巨头正向国会初选注资数亿美元作为“保险”，以防范地方监管并保护价值1万亿美元的基础设施，尽管选民因公用事业费用和数据中心问题爆发的基层反对浪潮正在高涨。面对华盛顿在立法层面的停滞不前，人工智能行业已向国会中期选举注入近3亿美元，旨在推选能够起草联邦监管规则的议员。该行业内部正因相互竞争的超级政治行动委员会（super PAC）而分裂：主张放宽监管的团体如Leading the Future（获1.4亿美元资金）极力推行宽松监管以在竞争中超越中国，而以安全为导向的派系如Anthropic支持的Public First Action（获8300万美元资金）则要求设立严格的安全防范措施。这些竞选活动采用了令人联想到传统烟草业和加密货币Fairshake PAC的激进说服策略，专门针对党内初选以撬动安全选区。然而，这场企业的政治闪电战正遭遇基层对当地数据中心建设的强烈抵制，同时前行业研究人员也对人类灭绝风险发出了严厉警告。与AI立场一致的政治行动委员会已筹集近3亿美元用于影响国会竞选，推动中期选举的政治广告总支出达到创纪录的110亿美元。 (0:01) AI游说团体已出现分裂：一方是主张放宽监管的超级政治行动委员会Leading the Future（拥有1.4亿美元），另一方是主张建立严格安全防范措施、由Anthropic支持的Public First Action（拥有8300万美元）。 (2:23) 随着科技公司承诺在数据中心和核心基础设施上投资约1万亿美元，数千万美元的超级政治行动委员会投入相当于一份确保有利监管的低成本保险。 (2:23) 由于公用事业费用和土地使用问题，基层对数据中心的抵制正持续升级，弗吉尼亚州一家上诉法院的一项裁决就叫停了一项占地近2000英亩、规划包含35个中心的项目。 (7:27)

Reading summary

Facing legislative inertia in Washington and rising public backlash over jobs and energy use, the artificial intelligence sector has poured nearly $300 million into midterm primary races to shape who writes future federal regulations. Historic Midterm Spending and Escalating Voter Anxiety (0:01) The midterm election cycle has surpassed $11 billion in total political advertising spending, making it the most expensive in U.S. history. As AI emerges as an issue cutting across traditional party lines, newly formed AI-aligned political action committees have raised close to $300 million to influence congressional races. This aggressive spending coincides with widespread voter unease regarding automation, job displacement, eroded personal autonomy, and the rapid rollout of energy-intensive computing infrastructure. Competing Super PAC Strategies in Primary Elections (2:23) Because gridlock stalls substantive legislation in Congress, the tech sector has focused on intervening in party primaries within safe districts to install favorable candidates before general elections begin. However, the AI lobby is split between two camps with opposing regulatory goals. The super PAC Leading the Future has raised $140 million to push for light-touch federal oversight that preempts state-level restrictions and prioritizes outpacing China. In response, Public First Action, backed by an estimated $83 million from Anthropic and led by Brad Carson, backs candidates from both parties who support strict safety standards and public guardrails. Persuasion Campaigns and the Corporate Influence Playbook (4:30) All capital deployed by these super PACs goes toward persuasion spending and media advertising, increasingly generated and targeted using AI systems. Analysts compare the lobby's tactics to the playbook pioneered by Big Tobacco, citing efforts to support weak federal frameworks, preempt tougher state laws, and evade legal liability. While the strategy mirrors the crypto industry's Fairshake super PAC—which used concentrated capital to defeat unfavorable politicians—AI advocates encounter far deeper resistance from the public due to tangible community impacts like grid strain, land use, and labor concerns. Local Data Center Resistance and Internal Whistleblowers (7:27) The political struggle over tech policy faces concrete friction at the municipal level, where residents are fighting sprawling data center projects that consume local water, increase electricity rates, and bulldoze woodlands. In Northern Virginia, community organizing led by Roger Yackel secured a unanimous Court of Appeals ruling voiding plans for more than 35 data centers spanning nearly 2,000 acres near Manassas National Battlefield Park. At the same time, internal division within the tech industry persists; former OpenAI and Anthropic researcher Jacob Coxon resigned in protest, warning that humanity is more likely than not to lose control of advanced AI systems under current trajectories. The central unresolved question is whether massive political spending will allow tech firms to dictate their own regulatory governance, or whether elected officials will retain enough independence to prioritize constituent concerns over corporate interests.

精炼阅读摘要

根据彭博原创（Bloomberg Originals）的报道，随着选民对就业替代、电网负荷及生存风险的焦虑升温，人工智能行业在破纪录的中期选举中投入近3亿美元政治资金，试图通过左右初选直接掌握未来科技监管规则的制定权。选民焦虑与破纪录的政治献金 (0:01) 本届中期选举成为美国历史上最昂贵的选举周期，全美政治广告支出突破110亿美元。与传统党派议题不同，人工智能跨越了党派界限，引发了关于白领与蓝领岗位流失、个人自主权削弱以及能源密集型数据中心扩张的普遍焦虑，甚至在OpenAI等企业总部外引发草根抗议。面对公众对科技精英单方面推行未经验证系统的反弹，华盛顿政界在推动对华竞争与设立安全护栏之间陷入分裂与立法停滞。为打破国会僵局并确保未来的监管杠杆掌握在支持本行业的议员手中，与AI行业结盟的超级政治行动委员会（Super PAC）筹集了近3亿美元。与科技行业单年在数据中心基础设施上投入约1万亿美元、并在核心研发上投入数千亿美元的规模相比，业内人士将这笔数亿美元的政治献金视为保护巨额资本投资、换取有利治理规则的低成本“保险”。两大超级PAC的初选代理人战争 (2:23) 面对难以通过实质性立法的国会，科技资本将策略从传统游说转向直接干预选举。其核心打法是在安全选区的党内初选中投入资金，在普选开始前直接选出符合自身政策倾向的候选人。然而，AI阵营内部在此分裂为理念截然对立的两大派系，各自提出完全相反的监管蓝图。由去监管阵营主导的超级PAC“Leading the Future”筹集了约1.4亿美元，主张推行轻度联邦监管，旨在压制各州碎片化的立法限制，并强调保持对中国的技术竞争优势。针锋相对的“Public First Action”则由Brad Carson领导，获得安全导向型AI实验室Anthropic约8300万美元的资助，该机构在两党初选中均支持主张严格安全标准与公共利益护栏的候选人，使初选成为科技行业争夺立法起草权的代理人战场。劝说广告策略与烟草业游说套路 (4:30) AI超级PAC将全部资金投入到劝说性广告中，讽刺的是，这些广告正是利用正在受争议的生成式AI系统进行快速制作和微定向投放，导致误导性或超逼真政治信息的制造成本大幅降低。分析人士指出，当前AI游说团体的修辞策略与昔日烟草巨头的手段高度相似：表面上呼吁联邦标准，实则抵制实质性执法机制；力推联邦优先权以废除更严格的州级法规；并在声称关注潜在危害的同时规避法律责任。这一政治运作模式在法律上依托于取消企业独立支出上限的“联合公民案”（Citizens United），并在策略上效仿了加密货币行业在2024年成立的超级PAC“Fairshake”。Fairshake曾通过地毯式广告轰炸成功清除了反对加密货币的政客。然而分析指出，AI行业面临着远比加密货币恶劣的舆论环境：加密货币主要局限于金融领域，而AI直接侵占物理社区资源、消耗紧缺公用事业，并威胁脑力劳动者的生计，难以仅靠巨额广告扭转民意。地方资源冲突与行业内部安全分歧 (7:27) AI监管的抽象争论在地方层面遭遇了剧烈的现实反弹。在原本规划为零售园区和绿地的郊区与乡村，居民们被迫与庞大的工业综合体和嗡嗡作响的变电站为邻。由于数据中心大量消耗区域水资源、推高居民电费并砍伐林地，社区反抗情绪高涨。在弗吉尼亚州北部，社区活动人士Roger Yackel联合居民发起抗议，成功促使上诉法院全票推翻了一项在马纳萨斯国家战场公园附近占地近2000英亩、拟建35个以上数据中心的大型项目。地方层面的强烈抵触凸显了PAC资金的局限性，数百万美元的广告难以平息电费激增和环境破坏带来的愤怒。与此同时，行业内部针对前沿模型的真实风险也出现分裂。尽管企业高管在听证会上坚称技术安全并承诺缓解能源成本，但顶尖安全研究人员正因担忧失去控制而相继辞职。曾在OpenAI和Anthropic任职的前研究员Jacob Coxon警告称，按照目前的发展轨迹，人类极有可能失去对这些AI的控制并面临灭绝风险，这让立法者深感由科技领袖起草自身监管法规无异于“引狼入室”。围绕人工智能治理的核心矛盾，本质上是企业利益与公共利益之间未决的权力博弈。随着史无前例的政治献金涌入中期选举，核心疑问在于：未来国家层面的AI监管规则究竟会被科技巨头的资本所左右，还是立法者能够顶住资金压力，真正依据选民对能耗、失业和生存安全的诉求来独立立法。

Detailed notes — zh

## Background 该视频由 YouTube 频道 Bloomberg Originals 于 2026 年 10 月 16 日发布。视频简介指出，人工智能已成为 2026 年美国中期选举的核心议题，行业利益集团斥资数亿美元支持有利于自身立法的候选人，而巨额政治资金正与公众对能源消耗、失业风险、数据中心扩张及生存威胁的深刻焦虑发生激烈碰撞。 ## TL;DR 在总支出超 110 亿美元、创下历史纪录的美国中期选举中，AI 产业正通过超级政治行动委员会（Super PAC）倾注巨资，争夺未来监管规则的主导权。以“引领未来”为代表的阵营斥资 1.4 亿美元推动联邦层面的宽松弱监管以抗衡国际竞争，而获得 Anthropic 等注资的“公共优先行动”则主张设立严格的安全护栏以保障公共利益。然而，科技巨头复制加密货币游说策略的企图遭遇了现实阻碍：地方社区对高能耗、砍伐森林的数据中心建设爆发了激烈抗议甚至提起诉讼，而行业内部对技术失控乃至人类灭绝风险的悲观警告，更让立法究竟服务于资本利益还是选民福祉成为了难以调和的核心拷问。 ## Atmosphere & style 全片充斥着调查性新闻的厚重质感与尖锐的批判审视，既有高层政治博弈的冷静解构，又有底层民众反弹的情绪张力。镜头在国会政治献金运作、巨型数据中心推土机的轰鸣、基层社区愤怒的抗议集会以及前核心研究员关于人类可能灭绝的灾难预警之间切换，形成强烈而刺目的反差。叙事语调兼具冷峻与讽刺，让观众在沉浸于快节奏事实呈现的同时，深切体会到在万亿级科技资本与无底洞般的政治广告轰炸面前，普通个体所面临的无力感与深刻焦虑。 ## Bulletin - [00:01](https://www.youtube.com/watch?v=wraPuftwsxQ&t=1s) 2026年美国中期选举政治广告支出超过110亿美元，刷新历史纪录，其中AI相关政治团体筹集近3亿美元用于争夺规则制定权。 - [02:23](https://www.youtube.com/watch?v=wraPuftwsxQ&t=143s) 两大AI超级政治行动委员会形成对立局面：“引领未来”筹集1.4亿美元推动宽松的联邦轻度监管，“公共优先行动”获Anthropic资助约8300万美元主张建立严格安全护栏。 - [07:45](https://www.youtube.com/watch?v=wraPuftwsxQ&t=465s) AI产业今年仅在数据中心基础设施上的投入就达到1万亿美元，巨额投资在基层社区引发严重邻避效应，包括电费上涨、噪音与水资源争夺等矛盾。 - [07:45](https://www.youtube.com/watch?v=wraPuftwsxQ&t=465s) 弗吉尼亚州上诉法院一致裁定驳回在马纳萨斯国家战场公园附近占地近2000英亩、拟建35座数据中心的项目，凸显基层民意对科技巨头扩张的强烈抵抗。 ## Key takeaways - [00:01](https://www.youtube.com/watch?v=wraPuftwsxQ&t=1s) 本届中期选举政治广告总支出超过110亿美元，成为美国历史上最昂贵的一次选举，AI超级政治行动委员会成为推动资金飙升的核心推手。 - [00:01](https://www.youtube.com/watch?v=wraPuftwsxQ&t=1s) 公众对AI的焦虑已跨越党派界限，主要集中在失业风险、个人自主权丧失以及科技精英自上而下强加颠覆性技术的不满。 - [00:01](https://www.youtube.com/watch?v=wraPuftwsxQ&t=1s) 加速AI落地阵营将技术发展上升至中美地缘政治竞争高度，认为谁赢得AI竞赛谁就掌握了全球竞争主导权。 - [02:23](https://www.youtube.com/watch?v=wraPuftwsxQ&t=143s) 科技企业改变游说策略，由推动国会立法转向直接通过巨额献金介入两党初选，以此扶植符合自身利益的代理人以掌控规则制定权。 - [02:23](https://www.youtube.com/watch?v=wraPuftwsxQ&t=143s) “引领未来”（Leading the Future）超级政治行动委员会筹资1.4亿美元，主张建立单一联邦弱监管框架，防止各州出台碎片化严苛规则。 - [02:23](https://www.youtube.com/watch?v=wraPuftwsxQ&t=143s) “公共优先行动”（Public First Action）获得Anthropic约8300万美元资助，其成立目的明确在于直接反击宽松监管阵营并力推安全护栏。 - [04:30](https://www.youtube.com/watch?v=wraPuftwsxQ&t=270s) AI行业游说借鉴了烟草巨头历史策略——以表面支持统一联邦监管为掩护，实则规避实质性法律责任与各州严厉监管。 - [04:30](https://www.youtube.com/watch?v=wraPuftwsxQ&t=270s) “联合公民案”（Citizens United）判决彻底解除了企业政治献金限制，为科技资本通过资金优势左右选战铺平了制度道路。 - [04:30](https://www.youtube.com/watch?v=wraPuftwsxQ&t=270s) 相比加密货币行业2024年的政治动员，AI游说难以轻易复制其路径，因AI深刻影响普通人日常生活并引发更广泛的公众关注与疑虑。 - [07:45](https://www.youtube.com/watch?v=wraPuftwsxQ&t=465s) 大规模数据中心建设在全美城镇引发严重邻避效应，破坏森林植被、带来高能耗与高噪音，并导致当地居民电费支出被动上涨。 - [07:45](https://www.youtube.com/watch?v=wraPuftwsxQ&t=465s) 弗吉尼亚州法院推翻2000英亩数据中心项目判决表明，基层民众对生态和生活环境受损的反弹，难以仅靠巨额政治游说资金平息。 - [07:45](https://www.youtube.com/watch?v=wraPuftwsxQ&t=465s) AI产业内部就存在主义风险产生剧烈分歧，前核心技术研究员公开警告技术失控可能导致人类走向灭绝风险。 - [10:46](https://www.youtube.com/watch?v=wraPuftwsxQ&t=646s) 未来AI治理的核心博弈在于：监管规则究竟是由科技资本的巨额献金主导，还是由民选官员代表选民的实际利益来确立。 ## Detailed walkthrough ### [00:01](https://www.youtube.com/watch?v=wraPuftwsxQ&t=1s) AI热潮下的选民焦虑与史上最昂贵选举在美国的中期选举中，选民关心的议题向来错综复杂，但人工智能（AI）正以前所未有的态势跨越党派界限，牵动着经济、环境乃至人类未来的核心命脉。伴随着全社会对“AI前景究竟有多宏大”的铺天盖地宣传，普通民众的焦虑感却在持续蔓延。一次又一次的民意调查清楚显示，大众对这项技术对其未来生活的冲击充满了不安：人们既害怕失去工作、丧失自主权，更存在一种深层的存在主义危机感——担心精英阶层正强行将颠覆性技术强加给普通人。这种民间恐慌情绪恰好与席卷此次选举的民粹主义暗流交织碰撞。在 OpenAI 办公室门外，抗议者甚至公开举牌怒斥：“AI游说集团正在买卖我们的政客！” 面对AI的迅猛发展，政界与产业界的领导力却呈现出割裂与涣散的状态。关于行业自律还是填补监管真空的争论不休，支持加速技术落地的一方将此上升到地缘政治对抗的高度。一位支持竞争的发言者强调：“我们在AI领域领先于中国，坦率地说，我希望保持这种状态，因为谁赢得了AI，谁就赢了。”（“We're leading China in AI, and frankly, I want to keep it that way, because whoever wins AI wins.”）然而，从各国政府呼吁建立国际标准的全球合作，到科技巨头高管与国家元首们对未来走向的茫然，没有人能准确预测AI的终局。这场围绕“谁来制定AI游戏规则”的权力博弈，正深刻地重塑美国人选举领导人的方式。此次中期选举不仅刷新了中期选举的支出纪录，更演变为美国历史上耗资最昂贵的一届选举，仅政治广告投入就远超110亿美元。推动这一吸金狂潮的关键引擎正是新兴的AI超级政治行动委员会（Super PAC），它们正将数以亿计的资金倾泻进初选阶段。面对大企业以无上限政治献金压制普通选民声音的现实，报道中甚至带着讽刺意味调侃道：“大家别慌，美国民主绝对能在AI竞选这个无底资金黑洞中幸存下来，对吧？” - 本届选举广告支出总额远超110亿美元，成为美国历史上最昂贵的一次选举。 - 与AI立场相关的政治团体在中期选举中已筹集了近3亿美元资金。 - 选民对AI的焦虑集中在就业岗位被取代、个人自主权丧失，以及精英阶层自上而下强加技术控制。 --- ### [02:23](https://www.youtube.com/watch?v=wraPuftwsxQ&t=143s) 规则制定权之争：两大AI超级政治行动委员会的暗战科技企业逐渐意识到，想要推动国会通过有利的法案极其困难，因此他们改变了策略，直接转向选举环节——试图通过重金扶植同情自身诉求的国会议员来左右政策。表面上看，这依然是一场政府与行业之间经典的“支持监管 vs 呼吁去监管”的拉锯战，两大掷下重金的阵营也都自称“支持AI发展”，但只要掀开表象就会发现截然相反的底色：一方极力寻求极为宽松的联邦轻度监管标准，另一方则要求设立严格、全面的AI安全护栏。在宽松监管一方，名为“引领未来”（Leading the Future）的超级政治行动委员会高调宣布筹集了1.4亿美元，专门资助反对严苛监管与安全护栏的亲AI候选人。该阵营主张建立全国性的单一监管框架，强烈反对各州自行其是出台碎片化的地方法规，并警告称如果美国不尽快奠定统一基础，“中国不仅会在竞赛中胜出，还会主导游戏规则”。这种大规模吸金举动随即引发了安全阵营的强烈反弹。主张对AI技术施加严格安全护栏的阵营迅速创立了与之抗衡的超级政治行动委员会——“公共优先行动”（Public First Action）。该组织获得了AI巨头 Anthropic 约8300万美元的巨额注资支持。 “公共优先行动”负责人布拉德·卡森（Brad Carson）明确阐述了该组织的创立初衷：“我们成立‘公共优先行动’组织，就是为了直接反击‘引领未来’。”（“Our group Public First Action was a direct response to try to fight back against Leading the Future.”）卡森指出，他们活跃在全美两党的各项选举中，目标是确保那些坚信“AI必须符合公众利益、而非损害公众利益”的候选人赢得初选。在政治运作逻辑中，介入各党初选并在安全选区选出顺从自身诉求的候选人，是对国会施加影响最立竿见影的手段。对于一个仅在今年就向数据中心建设投入1万亿美元、并向相关上下游企业注入数千亿美元的庞大产业而言，拿出1亿美元用于政治捐款以保护其庞大资产，在商业回报率（ROI）考量下不过是九牛一毛。 ```mermaid flowchart TD accTitle: AI超级政治行动委员会对立格局 accDescr: 对比Leading the Future与Public First Action在资金支持、政策主张与核心目标上的分歧 subgraph LTF["Leading the Future 阵营"] A1["募资规模：约1.4亿美元"] A2["主张：轻度联邦标准 / 反对全面护栏"] A3["核心论点：防范各州碎片化监管，抗衡中国竞争"] end subgraph PFA["Public First Action 阵营"] B1["关键资助：Anthropic 支持约8300万美元"] B2["主张：严格安全护栏 / 强化技术风控"] B3["核心论点：AI必须服务于公众利益而非资本利益"] end subgraph Goal["选举行动目标"] C1["大规模介入两党国会初选"] C2["锁定安全选区扶植代理人"] C3["主导未来国会AI立法与规则制定权"] end LTF --> C1 PFA --> C1 C1 --> C2 --> C3 ``` - AI行业今年仅在数据中心基础设施上的建设支出就达到1万亿美元。 - “引领未来”（Leading the Future）筹集资金达1.4亿美元，主张联邦单一弱监管。 - “公共优先行动”（Public First Action）获得 Anthropic 等资金支持达8300万美元，由布拉德·卡森领导，主张全面安全防护。 --- ### [04:30](https://www.youtube.com/watch?v=wraPuftwsxQ&t=270s) 政治游说与说服：从烟草巨头、加密货币到AI的游说策略演变在政治竞选中，超级政治行动委员会几乎将100%的资金都砸向了广告投放，这在政治学上被称为“说服性支出”（persuasion spending）。讽刺的是，如今用于在选战中说服选民的工具，正是AI技术本身——它使得制作真假难辨的政治宣传内容变得空前廉价和简单。然而，政治领域金钱驱动的说服术并非新鲜事，烟草巨头早在几十年前就为这种游说公关树立了范本：表面上宣称支持联邦监管以对抗各州的严厉法则，实际上暗中阻挠一切实质性的联邦监管法案；嘴上高喊不希望年轻人吸烟，暗地里却鼓励青少年吸烟；口口声声承诺会对产品负责，却竭尽全力逃避所有法律责任。这种政治撒钱的实质从来不是向大众证明该行业有多么崇高，而是向政客们展示其能够左右选举结果的金钱支配力。最高法院具有里程碑意义的“联合公民案”（Citizens United）判决彻底打破了企业政治献金的上限闸门，导致此后的每一届选举都比前一届更加昂贵。2024年的大选不仅刷新了开销历史，更为科技行业上了一堂将财富转化为政治权力的实操课。当时，加密货币行业创立了超级政治行动委员会“Fairshake”，筹集了惊人规模的资金并大肆挥霍。科技行业由此开始深谙华盛顿的政治游戏规则，通过“铺天盖地投放广告（flood the zone）、猛烈攻击异见候选人并将其拉下马”的打法，成功将加密货币议题推上国家政策前沿，甚至让政界喊出“美国必须成为全球比特币超级大国”的口号。然而，AI游说团体试图全盘复制加密货币的政治策略时却遭遇了瓶颈。加密货币的游说之所以容易推进，是因为全行业在其诉求的去监管方向上高度一致，且该技术本身远离大多数普通人的日常生活；但AI却完全不同——全美民众对AI的关注度与切身感受远超加密货币。AI正在直接改变普通人的社区面貌和物理基础设施，且对该技术的态度并不受传统民主党与共和党的意识形态界限约束。一位行业观察人士直言：“他们没有意识到，美国民众对AI的在乎程度远胜于当年的加密货币。”（“They didn't recognize that the American people care a lot more about AI than they ever did about crypto.”）对于AI企业而言，更大的挑战在于如何说服公众相信：这项技术不仅会使极少数科技精英受益，还能真正造福所有普通人。 - 政治行动委员会的资金几乎全部流向“说服性广告”，目前生成式AI自身已被广泛用于制造竞选内容。 - 烟草行业经典游说策略：表面呼吁统一的联邦法规以消解各州监管，实则全盘抗拒监管并逃避产品责任。 - 2024年加密货币行业通过“Fairshake”超级政治行动委员会展示了政治献金的杀伤力，但AI的受众面更广，无法简单复制其单一利益路线。 --- ### [07:45](https://www.youtube.com/watch?v=wraPuftwsxQ&t=465s) 数据中心落地引发的地方反弹与监管困境当AI公司从理论走向现实、开着推土机深入美国各地的普通城镇破土动工兴建数据中心时，引发了远超预期的巨大民意反弹。这种技术给社区带来的邻避效应（NIMBYism）极其尖锐：原先静谧的林地被砍伐一空，紧邻着联排居民住宅的空地上拔地而起的是冰冷庞大的数据中心和令人望而生畏的巨型变电站。许多购房者愤慨地表示，他们当初买房时收到的承诺是这里将建造便利的生活购物中心，结果等来的却是高噪音、高耗水且抬高居民电费账单的数据中心怪兽。地方居民对这种侵入式建设展开了顽强抵抗。在弗吉尼亚州马纳萨斯国家战场公园（Manassas National Battlefield Park）附近，科技企业曾计划在近2000英亩的土地上兴建超过35座数据中心。当地居民罗杰·亚克尔（Roger Yackel）等人联合发起诉讼并一路抗争，最终弗吉尼亚州上诉法院一致裁定驳回该开发项目。面对科技巨头的扩张，愤懑的居民甚至怒吼：“把你们的数据中心搬到火星上去建吧，我们这里不需要这玩意！”（“Bring your data centers to Mars and put them in there. We don't need any more here.”）这种来自基层社区关于电力账单上涨、水资源被侵占和环境恶化的切身痛感，是无论AI超级政治行动委员会在政治风向上投入数亿美元还是数十亿美元都无法简单摆平的现实矛盾。与基层的强烈抵制并存的，是AI产业自身深重的撕裂感。尽管AI投资已被视为拉动美国经济增长的重要引擎，多位科技巨头CEO也誓言要解决技术风险并控制飙升的能源成本，但全美各州乃至地方层面对立法的施压与日俱增。即便是最基础的监管原则，也在立法者之间存在剧烈分歧。更致命的分裂来自AI行业内部对技术危险程度的判断：曾先后就职于 OpenAI 和 Anthropic 的研究员雅各布·考克森（Jacob Coxon）因极度担忧技术失控而选择辞职，并公开发出灾难预警：“在当前的演进道路上，我认为人类更有可能对这些AI失去控制，这可能最终导致人类走向灭绝。”（“On the current path, I think it is more likely than not that humanity loses control to these AIs and it could end in human extinction.”）当行业对于自身终局存在如此两极化的认知时，让AI企业深度主导立法进程无异于“让狐狸看守鸡舍”，关于AI究竟需要多大规模与何种形式的监管，已成为当代最棘手的政治与制度难题。 - 弗吉尼亚州马纳萨斯国家战场公园附近一项占地近2000英亩、规划超过35座数据中心的超级项目，因居民上诉被法院全票否决。 - 社区抗议的核心矛盾聚焦于巨型变电站入驻、森林植被被砍伐、地下水资源消耗以及本地居民家庭电费被动飙升。 - 业界对AI存在主义风险的内部争议加剧，前核心研究人员公开发出人类可能丧失控制权乃至遭遇物种灭绝的严厉警告。 ### [10:46](https://www.youtube.com/watch?v=wraPuftwsxQ&t=646s) 谁来书写未来规则：企业政治献金与选民民意的博弈在围绕人工智能治理的诸多议题中，最难以达成的平衡点在于：未来的立法规则究竟最有利于行业巨头，还是最有利于普通大众。这场围绕中期选举展开的庞大资本运作，其底层逻辑归根结底就是未来技术规则的制定权之争。核心矛盾直指美国政治生态的运作方式：科技大厂铺天盖地的政治献金是否将彻底主导并决定美国 AI 监管的未来走向？抑或是民选官员能够顶住资本游说的压力，真正基于选区选民的关切与诉求来确立自己的政策立场？正如讲述者在总结时深刻指出的核心拷问：“This corporate spending, is that going to be what determines the future of AI regulation in this country or are politicians going to be able to stake out their own positions based on what their constituents want?” 这场立法主导权的拉锯，直接关乎未来规则制定究竟是以科技资本的利润为导向，还是以公众的实际福祉为依归。 - **根本利益之争**：监管框架的最难平衡之处在于利益倾斜方向——是优先保障 AI 产业的增长与自由度，还是优先保护普通民众的权益。 - **决定力量的博弈**：主导未来 AI 法律走向的力量究竟是来自巨额的企业竞选资金，还是来自反映选民真实意愿的政治独立立场。 ## Notable quotes > [07:45](https://www.youtube.com/watch?v=wraPuftwsxQ&t=465s) On the current path, I think it is more likely than not that humanity loses control to these AIs and it could end in human extinction. — 雅各布·考克森 (Jacob Coxon) > [00:01](https://www.youtube.com/watch?v=wraPuftwsxQ&t=1s) We're leading China in AI, and frankly, I want to keep it that way, because whoever wins AI wins. — 一位支持竞争的发言者 ## Mentioned OpenAI, Anthropic, Leading the Future, Public First Action, Fairshake, Brad Carson, Roger Yackel, Jacob Coxon, Manassas National Battlefield Park, Citizens United ## Video description AI is a major force in the 2026 US midterm elections, with industry groups spending hundreds of millions of dollars to elect candidates who might write the rules governing the controversial technology. But this unprecedented flood of money is splashing against a backdrop of growing public anxiety around AI’s rising energy costs, potential job destruction, sprawling data centers and existential risk. -------- Like this video? Subscribe: http://www.youtube.com/Bloomberg?sub_confirmation=1 Get unlimited access to Bloomberg.com for just $1.99 your first month: https://www.bloomberg.com/subscriptions?in_source=YoutubeOriginals Bloomberg Originals offers bold takes for curious minds on today’s biggest topics. Hosted by experts covering stories you haven’t seen and viewpoints you haven’t heard, you’ll discover cinematic, data-led shows that investigate the intersection of business and culture. Exploring every angle of climate change, technology, finance, sports and beyond, Bloomberg Originals is business as you’ve never seen it. Subscribe for business news, but not as you've known it: exclusive interviews, fascinating profiles, data-driven analysis, and the latest in tech innovation from around the world. Visit our partner channel Bloomberg News for global news and insight in an instant. [Source](https://www.youtube.com/watch?v=wraPuftwsxQ)

Source quotes

[7:45] 雅各布·考克森 (Jacob Coxon): On the current path, I think it is more likely than not that humanity loses control to these AIs and it could end in human extinction. [0:01] 一位支持竞争的发言者: We're leading China in AI, and frankly, I want to keep it that way, because whoever wins AI wins.

Source quotes

[2:23] Brad Carson: The way that you really influence Congress is by intervening in primaries and electing candidates who will win safe districts who will be friendly to your agenda, [7:27] Jacob Coxon: On the current path, I think it is more likely than not that humanity loses control to these AIs and it could end in human extinction,

Original transcript

[0:01] There's always a lot on the minds of voters, [0:03] and this critical midterm is no exception. [0:06] But one issue is cutting across party lines, [0:09] touching upon the economy, environment, [0:11] and possibly the future of humanity. [0:15] All we've heard about is how big AI will be. [0:17] It's growing in the United States, at least for now, [0:20] is this nimbyism and this pushback [0:22] to data centers getting constructed. [0:23] Poll after poll after poll shows that there's a lot [0:26] of anxiety among the general population about AI [0:29] and what it's going to mean for their futures. [0:31] Fear of job loss, fear of a loss of autonomy. [0:35] There's these bigger existential anxieties about the elite [0:40] imposing technologies on the rest of us. [0:44] And it's coinciding with this populist undercurrent [0:47] that has defined these midterms. [0:49] This is outside the OpenAI offices. [0:52] AI lobby be buying and selling our politicians. [0:55] Leadership on the issue has been scattered [0:57] and divided. [0:58] Self-regulation. [0:59] Regulatory voids. [1:00] We're leading China in AI, [1:02] and frankly, I want to keep it that way, [1:03] because whoever wins AI wins. [1:05] This is the largest industrial revolution in history. [1:07] We all need to work together, [1:09] called for global cooperation across the world [1:12] between governments to set international standards. [1:15] Where is AI going to go? I don't know. [1:18] I don't think presidents and CEOs really know. [1:21] There's a fight right now over who writes the rules of AI. [1:25] It's a battle for the soul of the AI industry. [1:27] And this battle is having profound effects on [1:30] how Americans elect their leaders. [1:32] This midterm, not only being the most expensive midterm [1:35] ever, but the most expensive election ever in the United [1:38] States, well over $11 billion spent on advertising. [1:41] And part of that is being fueled by these new AI super PACs [1:45] that are spending heavily in primary elections. [1:48] AI aligned political groups have raised close [1:50] to $300 million for the midterms. [1:53] This is the unlimited sums [1:56] of money coming from corporate America, [1:58] potentially outweighing the voices [2:00] of everyday Americans in the political process. [2:03] Nobody panic. [2:03] American democracy will totally survive the [2:06] bottomless money pit of AI campaigning, right? [2:11] It's going to change elections forever. [2:23] What these companies have realized is it's very hard [2:26] to get legislation through Congress, so they're turned [2:28] to the electoral process to try to elect their own members [2:31] of Congress who will be sympathetic to their agenda. [2:35] In many ways, this is the classic pro-regulation versus [2:37] deregulation debate that government [2:39] and industry always have. [2:40] The two big money spenders [2:42] that we're talking about say they're pro-AI, [2:44] but if you look under the hood a little bit, [2:46] there's a big difference. [2:47] One side wants very light touch federal standards. [2:51] The other side wants sweeping AI safety guardrails. [2:56] One of these big money spenders is called Leading the [2:58] Future, which claimed to have raised $140 million [3:01] to back pro-AI candidates [3:03] who oppose sweeping regulations or guardrails. [3:07] We want a national regulatory framework, [3:09] but we can't address these issues in a piecemeal fashion at [3:12] the state level because if we don't create this foundation, [3:15] not only is China going to win the race, [3:17] but then they're going to be calling the plays. [3:20] Leading the Future's massive fundraising success triggered [3:23] the founding of a competing super PAC in favor [3:25] of stricter guardrails on the encroaching technology. [3:32] My name is Brad Carson. [3:35] Our group Public First Action was a direct response to try [3:37] to fight back against leading the future. [3:39] Public First Action has received significant backing from [3:42] AI giant Anthropic, with an estimated $83 million overall. [3:48] Artificial intelligence. We are active in both [3:50] Democratic and Republican races across the nation [3:52] where we want to make sure people who believe [3:54] that AI should work in the public interest [3:56] and not against the public interest are supported. [3:59] The way that you really influence Congress is [4:01] by intervening in primaries [4:04] and electing candidates who will win safe districts [4:07] who will be friendly to your agenda. [4:09] But for an industry that's putting a trillion dollars just [4:11] into building data centers this year [4:13] and probably hundreds of billions more into the various [4:15] companies that are working in [4:17] and around AI, 100 million is nothing [4:20] to safeguard that investment. [4:22] If they can create rules [4:23] that benefit their industry that's a huge return on investment. [4:30] Advertising is literally 100% of what they're spending [4:33] In politics, it's called persuasion spending. [4:36] And now the tools being used [4:38] to persuade voters include AI itself. [4:41] Making it cheaper and easier to produce content [4:44] that can be difficult to distinguish from reality. [4:47] But persuasion in politics, well, that's nothing new. [4:53] Big Tobacco set the template [4:55] where you say you want federal regulation [4:57] and no state regulation, [4:58] but then you fight all federal regulation. [5:00] What kind of a man is he? [5:01] You say you don't want young people to smoke, [5:03] but you quietly encourage young people to smoke. [5:06] You say that you'll take responsibility for your product, [5:08] but you do everything you can to avoid that liability. [5:11] You're not trying to persuade them of, of [5:13] how good your industry is. [5:14] You're trying to show them that you have power [5:16] to spend money and influence the outcome of their election. [5:20] And a lot more of this power was granted [5:22] with the Supreme Court's Citizens United ruling. [5:25] Unlimited amounts of money. [5:27] Which really opened the floodgates [5:28] to massive political spending. [5:30] There have long been anxieties about [5:32] what Citizens United would mean [5:34] for corporate influence, but we're really seeing it now. [5:38] Since Citizens United, [5:39] every single election now is more expensive than the last. [5:42] Indeed, 2024 was the most expensive election to date. [5:46] It was also the year that taught a nascent industry valuable [5:49] lessons in converting tremendous wealth into unprecedented [5:53] political influence. [5:54] The US has got to get clear rules, [5:56] and the only way that's going [5:57] to happen is if we elect representatives [6:00] that actually represent our values. [6:01] In 2024, the crypto industry started a super [6:05] PAC called Fairshake. [6:07] Got a huge amount of money from the crypto industry [6:10] and it spent big. [6:12] We saw the tech industry become more directly involved in [6:15] DC politics and start to kind of play the lobbying game [6:18] and realize that that does make an impact on the future [6:22] of regulation or of deregulation. [6:24] The United States will be the Bitcoin superpower [6:28] of the world and we'll get it done. [6:29] Flood the zone, attack candidates that you disagree with, [6:33] knock them out of office. [6:34] This is the crypto playbook. [6:37] In many ways, the AI lobbying community tried [6:39] to use this playbook, but AI is different. [6:43] They didn't recognize that the American people care a lot [6:46] more about AI than they ever did about crypto. [6:48] Crypto was easy because the industry was aligned on the [6:51] regulations that they want, [6:53] but the AI industry affects every single person. [6:56] It's affecting people's neighborhoods [6:58] and physical infrastructure in this country, [7:01] and there's no clear delineation [7:04] between Democrats and Republicans. [7:06] Everything is thrown out the window. [7:09] While lobbying can certainly make an impact, [7:11] there's a bigger challenge for these AI companies, which is [7:16] convincing the public that this technology is not just going [7:20] to improve the lives of a small group of people, [7:23] but for everyone. [7:24] And that's, that's kind of the bigger battleground. [7:27] In everyday towns, we are seeing these AI companies come in [7:31] and start to actually put shovels in the ground [7:34] and build out these data centers, [7:36] and that is proving to be much more controversial than some [7:39] initially thought. [7:40] We'll take a closer look at the pushback right after this. [7:45] This is an incredible comeback story. [7:48] Uncover more at bloomberg.com/videos. [7:57] This used to be woodlands. [8:00] Right next to these are town homes, [8:03] and the residents who bought the town homes were told [8:07] that this would all be like a shopping center. [8:10] Instead, they get data centers. Look at how close this is. [8:15] Data centers in people's backyards have aggravated them. [8:18] They are associated with environmental damage, water usage, [8:23] electricity bills going up. [8:25] This is a great example [8:26] of a monstrous electrical substation. [8:30] The pushback on data centers right now is growing. [8:35] Roger Yackel and other residents fought [8:37] to prevent a project [8:38] that would've included more than 35 data centers on nearly [8:41] 2,000 square acres [8:42] of land near Manassas National Battlefield Park. [8:46] It went to the Court of Appeals [8:49] and the Court of Appeals voted unanimously to say, [8:52] "We're throwing it out." So we had that success. [8:56] Bring your data centers to Mars and put them in there. [9:00] We don't need any more here. [9:02] Even if you spend hundreds of millions [9:04] or even billions of dollars on trying [9:06] to influence the political headwinds on AI, [9:11] if data centers are an unpopular topic [9:14] or if people are anxious about losing their jobs, [9:16] I don't know if money necessarily solves that problem. [9:20] At the same time, AI investment has become an important [9:23] driver of US economic growth. [9:24] Is your technology safe? [9:26] The industry has sought to be responsive [9:28] to public concern, in some cases urging federal regulation. [9:32] When we launch a new feature. [9:33] While CEOs pledge to address risks associated [9:36] with the technology and rein in soaring power costs. [9:41] But as AI's footprint grows, pressure for government action, [9:45] particularly at the state [9:46] and local level, is only increasing. [9:48] Even some of the most basic ideas out there [9:53] are controversial to a lot of lawmakers. [9:56] This is really going to be a long [9:58] protracted effort to find consensus. [10:02] And the industry itself is at odds over the best way [10:05] forward and the magnitude of the problems AI poses. [10:08] A former OpenAI and Anthropic researcher, Jacob Coxon, [10:12] quitting his job over his fear of where AI is going. [10:15] On the current path, I think it is more likely than not [10:19] that humanity loses control to these AIs [10:22] and it could end in human extinction. [10:24] Who's going to get to write the rules? [10:25] Who's going to have the seat at the table? [10:27] It's two very different visions of [10:28] how this industry should operate. [10:30] I think there is this concern with AI companies [10:32] that if you let the AI companies be too closely in charge [10:35] of setting the regulation, [10:37] that it's the fox guarding the henhouse, so to speak. [10:41] How much and what type [10:42] of regulations AI needs are incredibly difficult questions. [10:46] Perhaps the hardest balance to reach will be [10:48] who the rules benefit the most, the industry or the people. [10:53] What's at stake is who's going to write the rules [10:55] for the future, and that's really [10:57] what a lot of this comes down to. [10:59] This corporate spending, is that going to be [11:02] what determines the future of AI regulation in this country [11:06] or are politicians going to be able [11:08] to stake out their own positions based on [11:10] what their constituents want?