The Fox Guarding the Hen House: Why the White House's AI Framework Is a Blueprint for Regulatory Capture

And What We Can Do About It

This morning, the White House released “A National Policy Framework for Artificial Intelligence,” its legislative recommendations for how Congress should regulate AI. On first read, the four-page document appears measured and reasonable. It leads with protecting children, a priority few would argue against. It discusses safeguarding communities, respecting intellectual property, preventing censorship, enabling innovation, and educating Americans for an AI-ready workforce. The language is careful. The tone is balanced.

Here’s what the framework actually proposes, section by section:

Section I, Protecting Children and Empowering Parents: Build on the Take It Down Act. Give parents tools to manage privacy settings and screen time. Establish age-assurance requirements for AI platforms. Require features that reduce risks of sexual exploitation and self-harm to minors. Affirm that existing child privacy protections apply to AI. But also: avoid “ambiguous standards” and “open-ended liability” that could lead to litigation.

Section II, Safeguarding Communities: Protect residential ratepayers from increased electricity costs due to data centers. Streamline federal permitting for AI infrastructure. Combat AI-enabled fraud targeting seniors. Ensure national security agencies understand frontier AI capabilities. Provide grants and tax incentives for small businesses.

Section III, Intellectual Property: Let the courts decide whether training AI on copyrighted material constitutes fair use. Consider enabling licensing frameworks for rights holders to negotiate with AI providers. Protect individuals from unauthorized AI-generated replicas of their voice or likeness.

Section IV, Preventing Censorship: Prevent the government from coercing AI providers to ban or alter content based on partisan agendas. Provide Americans a means to seek redress if agencies try to censor expression on AI platforms.

Section V, Enabling Innovation: Establish regulatory sandboxes for AI applications. Make federal datasets accessible for AI training. And critically: “Congress should not create any new federal rulemaking body to regulate AI, and should instead support development and deployment of sector-specific AI applications through existing regulatory bodies with subject matter expertise and through industry-led standards.”

Section VI, Workforce Development: Incorporate AI training into existing education and workforce programs. Study workforce trends driven by AI. Support land-grant institutions in developing AI programs.

Section VII, Federal Preemption: And here’s where the framework reveals its true purpose. “Congress should preempt state AI laws that impose undue burdens.” States should not be permitted to regulate AI development. States should not penalize AI developers for third-party unlawful conduct involving their models. Preemption must ensure state laws don’t “act contrary to the United States’ national strategy to achieve global AI dominance.”

Read in sequence, the structure becomes clear. The framework opens with consensus positions that nobody opposes (protect kids, help small businesses, respect creators). It buries the radical deregulatory provisions at the end, where they’re less likely to draw scrutiny. By the time you reach Section VII, you’ve been lulled into thinking this is a reasonable document. It is not.

The framework arrives at a pivotal moment: the Strait of Hormuz crisis has sent oil prices soaring past $120 per barrel, exposing just how fragile our fossil fuel dependence really is. Meanwhile, AI companies are racing to build energy-hungry data centers across the country.

I’ve read the recommendations carefully. The framework protects AI companies from accountability while concentrating power in federal agencies that have been systematically stripped of expertise. It does little to protect the American public.

We desperately need common-sense AI regulations. Guardrails that include kill switches. Robust data and privacy controls with real liability for violations. Comprehensive protections for children from deepfakes, sexualized content, and algorithmic manipulation. The European Union has shown us what a responsible regulatory framework looks like. What we’re getting instead is regulatory capture dressed up as innovation policy.

The Illusion of Oversight

The framework’s central argument sounds reasonable enough: “Congress should not create any new federal rulemaking body to regulate AI, and should instead support development and deployment of sector-specific AI applications through existing regulatory bodies with subject matter expertise and through industry-led standards.”

Why not leverage knowledge already resident in federal agencies? Sounds reasonable, right? Here’s the problem: those experts no longer exist within those agencies.

Over the past year, the Department of Government Efficiency (DOGE) has orchestrated the departure of approximately 300,000 federal workers. Engineers, scientists, economists, subject matter experts across virtually every agency. The Environmental Protection Agency has seen staff cuts approaching 65%. The Department of Health and Human Services announced 20,000 position cuts. NASA eliminated its Office of the Chief Scientist entirely. The Centers for Disease Control lost critical staff, including the entire Childhood Lead Poisoning Prevention and Surveillance Branch.

Max Stier, CEO of the Partnership for Public Service, put it bluntly: these mass layoffs show “a failure on the part of this administration to understand the critical value that the breadth of government expertise provides.” The layoffs led to “large departures of subject-matter experts resulting in the loss of institutional knowledge and technical expertise from the federal government.”

So here’s the reality: the administration hollowed out agencies, and now says oversight should be left to those same agencies. That leaves us with no meaningful oversight at all. Deregulation by attrition.

But there’s another layer to this. Even if these agencies still had their experts, the Supreme Court has already rendered that expertise legally irrelevant. In June 2024, the Court issued its landmark decision in Loper Bright Enterprises v. Raimondo, overturning the 40-year-old Chevron doctrine. For four decades, Chevron deference required federal courts to defer to agency interpretations of ambiguous statutes, recognizing that the specialists within those agencies understood their regulatory domains better than generalist judges. The Supreme Court, in a 6-3 decision, threw that out. Chief Justice Roberts declared that courts must now exercise “independent judgment” on all questions of statutory interpretation. Agency expertise no longer carries legal weight. As Justice Kagan warned in her dissent, this ruling replaces “a rule of judicial humility” with “a rule of judicial hubris,” handing technical regulatory questions to “hundreds of federal judges who lack the expertise of agency personnel.”

So put it all together. Congress, under this framework, won’t regulate AI directly and will block states from doing so. Instead, oversight goes to federal agencies. But those agencies have been gutted of expertise by DOGE. And even if they still had experts, the Supreme Court has ruled that their interpretations don’t deserve deference anyway. The whole structure is a circular feedback loop designed to produce one outcome: AI companies “self-regulate,” with no meaningful external check on their activities or power. The conflicts of interest here benefit Trump and those who have invested in him and him in their ventures, even though the framework reads as somewhat reasonable on first pass.

Preemption: Silencing the States

The framework’s seventh and final section may be its most consequential: “Congress should preempt state AI laws that impose undue burdens to ensure a minimally burdensome national standard consistent with these recommendations, not fifty discordant ones.”

But who defines “unduly burdensome”? The gutted federal agencies populated with industry-friendly appointees, apparently. The framework goes further, declaring that “States should not be permitted to regulate AI development, because it is an inherently interstate phenomenon with key foreign policy and national security implications.”

This concern is not hypothetical. In December 2025, President Trump signed an executive order establishing an AI Litigation Task Force within the Department of Justice, specifically tasked with suing states whose AI laws the administration dislikes. The administration has already targeted California’s AI Transparency Act and Colorado’s algorithmic discrimination law. Both provide protections that don’t exist at the federal level.

Just days before this framework was released, a bipartisan coalition of 42 state attorneys general sent a letter to major AI companies urging improved safeguards for children and mitigation of harmful model outputs. These officials represent both red and blue states. They understand their constituents need protection. The framework’s answer? Block them from providing it.

The “patchwork” argument ignores reality. In the absence of meaningful federal action, states have been the only entities willing to step up. California’s SB 53 and New York’s RAISE Act established whistleblower protections and reporting requirements that didn’t exist at the federal level. Preempting these protections without replacing them with something stronger amounts to regulatory abandonment. The framework released today mentions a “minimally burdensome national standard consistent with these recommendations,” but there is no national standard and the framework says there should not be one.

Section 230 on Steroids: The Liability Shield

The framework’s approach to liability troubles me most. In the section on protecting children, the document states that “Congress should avoid setting ambiguous standards about permissible content, or open-ended liability, that could give rise to excessive litigation.” The presumption here is that litigation is bad in all cases.

The preemption section goes further: “States should not be permitted to penalize AI developers for a third party’s unlawful conduct involving their models.”

Brad Carson, who leads Americans for Responsible Innovation, called this approach “230 on testosterone.” He’s referring to Section 230 of the Communications Decency Act, which shields internet platforms from liability for user-generated content. That liability shield has enabled platforms to profit from harmful content while claiming they bear no responsibility to moderate or remove it. Now it appears that Congress is being asked to extend similar protections to AI companies.

Consider what this means in practice. An AI system generates a deepfake of your child. An AI-powered recommendation algorithm drives a teenager toward self-harm content. An AI chatbot manipulates a vulnerable adult into harmful decisions. Under this framework, the companies that built these systems, deployed them at scale, and profited from them face minimal consequences, if any.

The framework does include language about protecting children, establishing age-assurance requirements and building on the Take It Down Act that criminalizes nonconsensual deepfake imagery. But the liability limitations undermine these provisions. Rules without meaningful consequences for breaking them accomplish little.

And the presumption that litigation is inherently negative deserves scrutiny. Litigation is one of the most important tools we have in this country to police damaging corporate behavior. Without it, there are no remedies and no accountability. This is especially true right now, when all branches of government have been captured by one extreme end of one party. It is litigation, and litigation alone, that is forging ahead in resisting the breakdown of our democracy, one case at a time. Yes, litigation can be abused by unscrupulous plaintiffs and unethical attorneys. But it remains a vital tool in our democratic system, providing remedies and creating incentives for companies and the government to do the right thing and do no harm. Any assumption that litigation should be avoided, that causes of action should not be provided for, should be a nonstarter for all Americans who care about their rights, the rule of law and accountability.

The Myth That Regulation Kills Innovation

Defenders of this framework will argue that strong regulation would stifle American innovation and hand leadership in AI to China. Every major era of technological innovation in our country’s history contradicts this claim.

Requiring seatbelts, airbags, and emission controls made the automobile industry adapt and improve. Establishing the FDA pushed the pharmaceutical industry to develop rigorous safety testing that has produced life-saving medications. Financial reforms following 1929 and 2008 made the financial industry more stable and trustworthy.

Throughout American history, regulation has been necessary to curb corporate greed, prevent labor abuses, address antitrust and RICO violations, and protect our air, water, and environment. No human is immune from acting in their own best interest when facing a conflict of interest. Additionally, corporate executives have fiduciary duties to maximize shareholder value. Those duties often directly conflict with responsible AI regulation, because meaningful regulations increase costs.

We cannot trust companies to regulate themselves. We cannot trust their leaders and employees to prioritize public safety over profits. This isn’t cynicism. It’s the recognition of basic human nature that has informed every successful regulatory regime in American history.

The EU has proven this point decisively. The EU AI Act began enforcement in 2025, with full implementation by 2027. It establishes a comprehensive risk-based framework with administrative fines reaching 35 million euros or 7% of worldwide turnover for prohibited practices. European companies haven’t fled. Major players like Google and Microsoft have embedded EU-compliant transparency tools into their global product suites. Apple is building green-energy data centers. Responsible regulation creates trust, and trust enables sustainable growth, while preventing catastrophic harm.

Conflicts of Interest at Every Level

The conflicts of interest surrounding this framework are breathtaking in scope. David Sacks, the White House’s AI and Crypto Czar who helped develop these recommendations, maintains over 400 investments in tech firms with ties to AI through his venture firm, Craft Ventures. He divested from some holdings like Amazon and Meta. Government ethics expert Kathleen Clark of Washington University characterized his waivers as “sham ethics waivers” that “lack the kind of rigorous objective ethics analysis that would ensure that public policy is made for public benefit.”

Even Steve Bannon has raised concerns. “Right now, you have more regulations, ten times more regulations, to open a nail salon on Capitol Hill than you have into one of the most promising yet one of the most dangerous technologies ever invented,” Bannon told NPR. “My question to this group: Where’s the risk mitigation? I haven’t seen it.”

The administration’s relationships with AI companies extend far beyond advisory roles. The government has taken stakes in chipmakers, struck deals taking 15% cuts of Nvidia’s chip sales to China, and created programs like the “Tech Force” that directly partner with Amazon Web Services, Apple, Google, Microsoft, Nvidia, OpenAI, Oracle, and Palantir. When participants complete their government service, these companies have committed to consider them for employment.

The revolving door between government and industry isn’t unique to this administration, but the scale and brazenness are unprecedented. When the people writing AI policy stand to profit from weak AI regulation, weak regulation is what we get.

The Environmental Crisis We’re Ignoring

The framework calls for streamlining federal permitting for AI infrastructure construction and ensuring that “residential ratepayers do not experience increased electricity costs as a result of new AI data center construction.” Fine words. They ignore the fundamental problem: AI data centers are being built on fossil fuel infrastructure at precisely the moment when we should be divesting from fossil fuel dependence.

The numbers are staggering. According to the International Energy Agency, global electricity generation to supply data centers is projected to grow from 460 TWh in 2024 to over 1,000 TWh by 2030. In the United States and China, most electricity consumed by data centers comes from fossil fuels. Natural gas is currently the biggest source of electricity for U.S. data centers at over 40%, with coal at around 15%.

The current crisis in the Strait of Hormuz demonstrates exactly why this matters. Oil prices have surged past $120 per barrel. Analysts are calling it “the biggest energy crisis since the oil embargo in the 1970s.” Roughly 20% of global oil and natural gas passes through that strait. Iran has effectively closed the waterway, and hundreds of tankers sit idle on both sides. Iraq has had to shut down production in some of its largest oil fields because it has nowhere to store the oil it can’t export.

This is the fragility we’re building into the foundation of American AI infrastructure. If we let AI companies build their data centers on fossil fuel dependence now, they will later complain that changing course is too expensive and disruptive. We’ve seen this playbook before with every incumbent industry that profits from the status quo.

But energy is only part of the story. Data centers are devouring America’s freshwater supplies. A single large data center can consume up to five million gallons of water per day for cooling, equivalent to the needs of a town of 50,000 residents. The water doesn’t just flow through and return to the system; much of it evaporates, permanently depleting local aquifers and municipal supplies. A study by the Houston Advanced Research Center found that data centers in Texas alone will use 49 billion gallons of water in 2025, potentially rising to 399 billion gallons by 2030. And according to Cornell researchers, by 2030 the AI industry could drain 731 to 1,125 million cubic meters of water annually, equivalent to the household water usage of 6 to 10 million Americans.

The industry is making this worse by building in exactly the wrong places. More than 160 new AI data centers have been built in the past three years in areas with high competition for scarce water resources. Nevada, Arizona, Texas: these are states already facing water stress, and data center developers are tapping their aquifers anyway because water is cheap compared to real estate and power. In Northern Virginia, the densest concentration of data centers anywhere in the world, local officials are raising alarms about strain on water resources and infrastructure.

Then there’s the pollution. Data center backup generators run on diesel, releasing nitrogen oxides at 200 to 600 times the rate of natural gas plants producing equivalent energy. According to the Institute of Electrical and Electronics Engineers, data centers caused approximately $6 billion in public health damages from air pollution in 2023 alone. Researchers estimate between 190,000 and 300,000 asthma symptom cases and 370 to 595 premature deaths annually from data center pollution. One in five data centers are sited in communities already overburdened by environmental pollutants. The NAACP has launched a “Stop Dirty Data Centers” campaign because these facilities are disproportionately located in Black and low-income communities, where residents already suffer higher rates of respiratory illness.

In Memphis, Elon Musk’s xAI data center installed more than 30 natural gas turbines intended for daily use, not just backup power. Local residents and the NAACP filed a notice of intent to sue under the Clean Air Act, arguing the project could worsen already dangerous air quality in a city with high asthma rates. Communities from rural Pennsylvania to suburban Arizona to historically Black neighborhoods are pushing back against data center projects that bring few permanent jobs but outsized impacts: spiking electricity demands, strained water supplies, noise pollution, and industrial infrastructure forced through residential and agricultural areas.

Europe is taking a different path. European data centers are increasingly located in Scandinavia and France, where cheap and abundant renewable and nuclear power is available. Almost 50% of European power came from renewables in 2024. But it’s not just about location. The EU’s Energy Efficiency Directive requires data center operators to report key performance indicators including energy consumption, water usage, waste heat utilization, and renewable energy use. Germany’s Energy Efficiency Act requires data centers to purchase 50% of their electricity from unsubsidized renewable sources since 2024, rising to 100% by 2027. New data centers must reuse at least 10% of their waste heat, increasing to 20% for facilities opening after July 2028.

The EU is demonstrating that regulation drives innovation. Faced with requirements for renewable energy, waste heat recovery, and sustainability reporting, European data center operators are developing advanced liquid cooling systems that reduce water consumption, siting facilities to feed waste heat into district heating networks, and investing in on-site renewable generation. These technologies exist. They work. They’re becoming standard practice in Europe because regulations forced the industry to innovate instead of defaulting to fossil fuels and freshwater consumption.

Meanwhile, the $500 billion Stargate initiative in the United States is building massive data centers that will require five gigawatts each, more than the entire power demand of New Hampshire. There’s no requirement that this power come from clean sources, no mandate for water recycling or efficiency, no obligation to capture and reuse waste heat. The framework’s vague promise to protect residential electricity rates, without any mechanism for accountability or any requirement for renewable energy, accomplishes nothing.

Here again the conflict of interest cannot be ignored. Many people in the Executive branch and Congress stand to profit from continued fossil fuel dependence. And water is becoming a profit center too. Investment funds have over $100 billion ready to invest in private ownership of water rights. Private equity firms are buying up agricultural land in water-stressed regions primarily for the water entitlements, not the farming. Companies like California American Water and Pennsylvania American Water are acquiring municipal water systems across the country, with six acquisitions in Pennsylvania planned for 2025 alone. The privatization of water infrastructure is accelerating precisely as data centers increase their demands on scarce supplies.

If we require AI companies to build sustainable infrastructure from the start, they will do it. They have the capital and the technical capacity. But if we let them build dirty, they will fight any subsequent cleanup requirements tooth and nail, claiming the costs are too high and the disruption too great. This is the pattern of every polluting industry in American history. The only way to break it is to set the requirements now, before the infrastructure is locked in. The framework does the opposite: it streamlines permitting and preempts state environmental regulations, clearing the path for AI companies to externalize their energy costs, their water consumption, and their pollution onto the communities unfortunate enough to host their facilities.

What This Means for Startups and Small Businesses

The framework pays lip service to small businesses, calling for “grants, tax incentives, and technical assistance programs, to support wider deployment of AI tools across American industry.” The reality is that the regulatory environment this framework creates will devastate startups and mid-sized companies while entrenching the power of Big Tech.

Here’s why. When federal preemption eliminates state-level consumer protections, the biggest players can absorb any resulting lawsuits and reputational damage. They have armies of lawyers and deep pockets for settlements. Small companies don’t. When something goes wrong with AI (and things will go wrong), small companies will be destroyed while big companies pay a fine and move on.

The concentration of AI infrastructure in the hands of a few massive players creates dangerous dependency. When you build your business on OpenAI’s API or Google’s cloud, you’re subject to their pricing, their terms of service, their decisions about what models to deprecate and when. The framework does nothing to address this power imbalance. By limiting liability and preempting state regulation, it makes the situation worse.

If you’re a startup or mid-sized company adopting AI right now, here’s what you should be doing:

Build to the highest standard. The EU AI Act has extraterritorial scope. If your AI affects EU users, you’re subject to EU rules regardless of where you’re based. Build your systems to EU standards now, and you’ll be prepared for whatever regulatory environment eventually emerges in the U.S.

Document everything. Keep records of your training data sources, your testing procedures, your risk assessments. When regulation eventually comes (and it will come), companies that can demonstrate responsible practices will be in a far better position than those that can’t.

Diversify your dependencies. Don’t build your entire business on a single AI provider. The concentration of power in Big Tech is a business risk, not just a policy concern.

Engage politically. Join industry associations that advocate for balanced regulation. Support state-level initiatives that protect consumers and create clear rules of the road. The absence of regulation benefits big business that can operate in legal gray zones. It does not benefit small business.

What Real AI Regulation Should Look Like

The EU has provided a template. The EU AI Act takes a risk-based approach, categorizing AI systems into four risk levels: unacceptable, high, limited, and minimal risk. Systems that pose unacceptable risks, like social scoring or untargeted facial scraping, are banned outright. High-risk systems in areas like healthcare, education, and employment must undergo pre-deployment assessments, maintain extensive documentation, implement human oversight, and conduct post-market monitoring.

This is common sense. We require impact assessments for environmental projects. We require safety testing for pharmaceuticals. We require crash testing for automobiles. AI systems that make decisions affecting people’s lives, livelihoods, and liberties deserve similar scrutiny.

Real AI regulation should include:

Kill switches and human override capabilities for AI systems operating in high-stakes domains. When an AI system is making decisions about healthcare, criminal justice, weapons and warfare, or critical infrastructure, humans must be able to intervene.

Robust data and privacy controls with real liability for violations. Companies should not be able to hoover up personal data for AI training without meaningful consent, and they should face serious consequences when they violate privacy rules, like is mandated in the EU.

Comprehensive protections for children that go beyond the Take It Down Act. This means not just criminalizing deepfakes after the fact, but requiring AI platforms to implement proactive measures to prevent the creation and distribution of exploitative content involving minors. It means protecting children from algorithmic manipulation designed to maximize engagement at the expense of their wellbeing.

Transparency requirements so people know when they’re interacting with AI and can understand how AI systems are making decisions that affect them. All AI videos, audios and images should be required to be labeled as such, which stiff penalties for a failure to comply.

Meaningful liability for AI developers when their systems cause harm. The shield provided by limiting liability to “third party unlawful conduct” is a massive loophole that will swallow any consumer protection.

Real copyright protection that goes beyond the framework’s vague gesture toward letting courts sort it out. The framework’s lip service to intellectual property does nothing to address the fundamental problem: AI companies have scraped up billions of copyrighted works without notice or authorization to train their models, operating on the Silicon Valley principle of moving fast and asking for forgiveness later. More than 90 copyright infringement lawsuits have now been filed against AI companies, including cases brought by the New York Times, Disney, the Authors Guild, visual artists, musicians, and software developers. These lawsuits will take years to resolve, and in the meantime the damage is done. Without regulations that uphold copyrights in the U.S., this behavior will effectively kill copyright protection. If there are no consequences for scraping the entire internet without permission, then copyright becomes meaningless.

We need either a mandatory licensing regime that requires AI companies to negotiate with rights holders before using their content for training, or meaningful liability for companies that fail to seek authorization. The argument that such requirements would hinder innovation falls as flat as every other argument against AI regulation. If these requirements are put in place now, the companies will find a way to innovate around them. Some already are: major publishers have signed licensing agreements with AI companies, demonstrating that the market can work when companies are incentivized to participate. But the incentive has to come from regulation, not voluntary goodwill. The EU AI Act requires AI providers to document their use of copyrighted content and comply with copyright law, with penalties up to €15 million or 3% of global turnover for violations. In contrast, the White House framework offers no enforcement mechanism whatsoever.

Sustainable infrastructure requirements for AI data centers, including mandatory use of renewable energy, water recycling and efficiency standards, waste heat recovery obligations, and binding pollution limits. The EU has shown this works. Germany requires 100% renewable electricity for data centers by 2027 and mandates waste heat reuse. If European operators can meet these standards, American operators can too, especially because often the European operators ARE the American operators, such as Apple, Google and Meta. Additionally, the costs of energy consumption, water depletion, and environmental degradation should fall on AI developers, not on communities and ratepayers.

The Stakes Could Not Be Higher

We are at an inflection point. The decisions made in the next few years about AI regulation will shape our society for decades. The White House’s framework represents a choice to prioritize corporate profits and rapid deployment over public safety and democratic accountability.

Other choices exist. We can demand common-sense regulation that protects children, ensures privacy, establishes liability, and requires sustainable infrastructure. We can support state-level initiatives that fill the gaps left by federal inaction. We can hold our elected officials accountable for their conflicts of interest and their cozy relationships with the industries they’re supposed to oversee.

The administration argues that strong regulation would hand AI leadership to China. But leadership built on exploitation of workers, consumers, and the environment isn’t leadership worth having. Real leadership means showing the world that technological progress and human values can coexist. That innovation and accountability can work together.

The fox cannot guard the hen house. It’s time to demand real oversight.

Sources and References

Primary Source Document

White House, “A National Policy Framework for Artificial Intelligence” (March 20, 2026)

White House AI Policy and Executive Orders

White House, “Fact Sheet: President Donald J. Trump Ensures a National Policy Framework for Artificial Intelligence” (December 2025)

https://www.whitehouse.gov/fact-sheets/2025/12/fact-sheet-president-donald-j-trump-ensures-a-national-policy-framework-for-artificial-intelligence/

White House Executive Order, “Eliminating State Law Obstruction of National Artificial Intelligence Policy” (December 2025) — Source for DOJ AI Litigation Task Force

https://www.whitehouse.gov/presidential-actions/2025/12/eliminating-state-law-obstruction-of-national-artificial-intelligence-policy/

DOGE and Federal Workforce Cuts

Wikipedia, “2025 United States federal mass layoffs” — Source for ~300,000 announced layoffs, 200,000+ departed by August 2025

https://en.wikipedia.org/wiki/2025_United_States_federal_mass_layoffs

Challenger, Gray & Christmas, “Federal Cuts Dominate March 2025”

https://www.challengergray.com/blog/federal-cuts-dominate-march-2025-total-275240-announced-job-cuts-216670-from-doge-actions/

GovExec, “Project 2025 wanted to hobble federal workforce. DOGE has hastily done that and more” (April 2025)

https://www.govexec.com/transition/2025/04/project-2025-wanted-hobble-federal-workforce-doge-has-hastily-done-and-more/404390/

Loper Bright Decision (Chevron Deference)

Supreme Court of the United States, Loper Bright Enterprises v. Raimondo, 603 U.S. ___ (June 28, 2024)

https://www.supremecourt.gov/opinions/23pdf/22-451_7m58.pdf

SCOTUSblog, “Supreme Court strikes down Chevron, curtailing power of federal agencies” (June 2024)

https://www.scotusblog.com/2024/06/supreme-court-strikes-down-chevron-curtailing-power-of-federal-agencies/

Conflicts of Interest

NPR, “David Sacks AI advisor investment conflicts” (December 2025) — Source for Craft Ventures 400+ AI investments, Kathleen Clark “sham” ethics waiver quote

https://www.npr.org/2025/12/12/nx-s1-5631823/david-sacks-ai-advisor-investment-conflicts

CNBC, “Trump AI Tech Force” (December 2025) — Source for Tech Force program with Amazon, Apple, Google, Microsoft, Nvidia, OpenAI, Oracle, Palantir

https://www.cnbc.com/2025/12/15/trump-ai-tech-force-amazon-apple.html

Foreign Policy, “Nvidia Chips, National Security, Trump, China, AI” (December 2025) — Source for 15% cut of Nvidia China sales

https://foreignpolicy.com/2025/12/12/nvidia-chips-national-security-trump-united-states-china-ai/

EU AI Act

Greenberg Traurig, “EU AI Act: Key Compliance Considerations” (July 2025) — Source for penalties up to €35 million or 7% of global turnover

https://www.gtlaw.com/en/insights/2025/7/eu-ai-act-key-compliance-considerations-ahead-of-august-2025

DLA Piper, “Latest wave of obligations under the EU AI Act take effect” (August 2025)

https://www.dlapiper.com/en-us/insights/publications/2025/08/latest-wave-of-obligations-under-the-eu-ai-act-take-effect

Energy and Data Centers

International Energy Agency, “Energy and AI” (April 2025) — Source for data center electricity: 460 TWh (2024) → 1,000+ TWh by 2030

https://www.iea.org/reports/energy-and-ai/energy-supply-for-ai

CNBC, “AI Energy Transition Fossil Fuels” (January 2026) — Source for U.S. data centers: 40%+ natural gas, ~15% coal

https://www.cnbc.com/2026/01/22/ai-energy-transition-fossil-fuels.html

Water Consumption and Environmental Impacts

Lincoln Institute of Land Policy, “Data Drain: The Land and Water Impacts of the AI Boom” (October 2025) — Source for Texas water use projections, 5 million gallons/day

https://www.lincolninst.edu/publications/land-lines-magazine/articles/land-water-impacts-data-centers/

Cornell Chronicle, “Roadmap shows the environmental impact of AI data center boom” (November 2025) — Source for 731-1,125 million cubic meters water annually by 2030

https://news.cornell.edu/stories/2025/11/roadmap-shows-environmental-impact-ai-data-center-boom

Bloomberg, “The AI Boom Is Draining Water From the Areas That Need It Most” (May 2025) — Source for 160+ new data centers in water-stressed areas

https://www.bloomberg.com/graphics/2025-ai-impacts-data-centers-water-data/

Brookings Institution, “AI, data centers, and water” (November 2025) — Source for 300,000 gallons/day typical; up to 5 million gallons/day for large facilities

https://www.brookings.edu/articles/ai-data-centers-and-water/

Air Pollution and Health Impacts

Environmental Data and Governance Initiative, “Communities Close to EPA-Regulated Data Centers Face Heightened Air Pollution” (October 2025)

https://envirodatagov.org/blogs/communities-close-to-epa-regulated-data-centers-face-heightened-air-pollution/

North Carolina Environmental Justice Network, “Resources on Data Centers” (August 2025) — Source for $5.7-9.2 billion annual health costs; 190,000-300,000 asthma cases; 370-595 premature deaths

https://ncejn.org/resources-on-understanding-data-centers-and-their-environmental-impacts/

UAB Institute for Human Rights, “The Human Impacts of AI Data Centers” (October 2025) — Source for IEEE study: $6 billion in health damages from air pollution in 2023

https://sites.uab.edu/humanrights/2025/10/02/construction-and-consequences-the-human-impacts-of-artificial-intelligence-data-centers/

World Resources Institute, “From Energy Use to Air Quality” — Source for xAI Memphis data center: 30+ natural gas turbines; NAACP lawsuit

https://www.wri.org/insights/us-data-center-growth-impacts

NAACP, “Stop Dirty Data Centers” Campaign

https://naacp.org/campaigns/stop-dirty-data-centers

European Data Center Regulations

European Commission, “Energy performance of data centres”

https://energy.ec.europa.eu/topics/energy-efficiency/energy-efficiency-targets-directive-and-rules/energy-efficiency-directive/energy-performance-data-centres_en

Data Center Group, “Energy Efficiency Act” (Germany) — Source for 50% renewable by 2024, 100% by 2027; waste heat reuse requirements

https://datacenter-group.com/en/company/sustainability/energy-efficiency-act/

Water Privatization

Flow Water Advocates, “Privatization and Commodification” — Source for investment funds: $100+ billion ready to invest in private water rights

https://flowwateradvocates.org/issues/privatization/

City & State Pennsylvania, “Private acquisition of municipal water system” (December 2025) — Source for Pennsylvania American Water: 6 acquisitions planned for 2025

https://www.cityandstatepa.com/policy/2025/12/private-acquisition-municipal-water-system-under-new-spotlight-utility-giants-plan-merge/409833/

Strait of Hormuz Crisis

NPR, “Iran war, oil, Strait Hormuz closed, energy crisis” (March 2026) — Source for “biggest energy crisis since the oil embargo in the 1970s”

https://www.npr.org/2026/03/04/nx-s1-5736104/iran-war-oil-trump-israel-strait-hormuz-closed-energy-crisis

CNBC, “Strait of Hormuz closure” (March 2026) — Source for 20% of global oil and natural gas passes through strait; oil past $120/barrel

https://www.cnbc.com/2026/03/03/strait-of-hormuz-closure-which-countries-will-be-hit-the-most.html

Copyright and AI Training

Copyright Alliance, “AI Copyright Lawsuit Developments in 2025” (January 2026) — Source for 90+ lawsuits; $1.5 billion Bartz v. Anthropic settlement

https://chatgptiseatingtheworld.com/2026/03/16/latest-us-map-of-copyright-suits-v-ai-cos-total-91-mar-15-2026/

https://copyrightalliance.org/ai-copyright-lawsuit-developments-2025/

Skadden, “Copyright Office Weighs In on AI Training and Fair Use” (May 2025)

https://www.skadden.com/insights/publications/2025/05/copyright-office-report

Best Law Firms, “AI’s War in the Courtroom: Copyright Disputes Spike in 2025” (December 2025) — Source for 50+ lawsuits pending in federal courts

https://www.bestlawfirms.com/articles/ai-war-in-the-courtroom-copyright-disputes-spike-in-2025/7186

Congressional Research Service, “Generative Artificial Intelligence and Copyright Law”

https://www.congress.gov/crs-product/LSB10922