Every Commission Studying AI Job Loss Is Funded by the Companies Causing It

In October 2025, NVIDIA and the Special Competitive Studies Project announced a Task Force on AI and the Future of Work. Jensen Huang, whose company sells the GPUs powering every major AI system, would co-direct it alongside Eric Schmidt's think tank. The task force would study what happens to workers displaced by the technology NVIDIA manufactures and sells.

Five months later, on March 11, 2026, Senators Mark Warner and Mike Rounds introduced the Economy of the Future Commission Act. Its endorsement list reads like an AI industry trade group: Google, Meta, Microsoft, IBM, Workday. The commission would study "workforce changes driven by artificial intelligence" — changes driven, in no small part, by the endorsers' own products.

Meanwhile, Congress passed the Workforce Pell expansion, subsidizing 8-to-15-week training certificates so displaced workers can retrain into fields that may not exist by the time they finish.

Three initiatives. Zero displaced workers at the table. A pattern economists have a name for: regulatory capture.

The Fox Designs the Henhouse

Regulatory capture usually describes a slow process. An industry funds research that happens to support its interests. Former executives rotate into government oversight positions. Lobbying shifts the scope of regulation toward the comfortable and away from the threatening. It typically takes decades.

With AI workforce policy, it happened in months.

Start with the SCSP-NVIDIA task force. SCSP was founded by Eric Schmidt in 2021 after the National Security Commission on AI — which Schmidt also chaired — concluded its work. SCSP describes itself as nonpartisan. Its stated mission is to "strengthen America's long-term competitiveness" in AI. Its work has focused heavily on military AI adoption and beating China. It collaborated with RAND on Taiwan invasion wargames. It pushed for a Pentagon experimentation unit to speed up generative AI adoption.

This is the organization studying what happens to the workers AI replaces.

The task force will deliver an interim report at SCSP's AI Expo in May 2026 and a final report in October. Its members are drawn from "industry, academia, and government." The NVIDIA press release mentions workforce development, national competitiveness, and talent cultivation. It does not mention levies, displacement measurement, transition support, or income replacement. The framing is: how do we skill up workers to serve the AI economy — not whether the AI economy needs guardrails that might slow NVIDIA's customers down.

The Endorsement List Is the Tell

The Warner-Rounds commission is more interesting because it's legislation, not a think-tank project. It would include bipartisan members of Congress alongside experts from industry and academia. Its scope explicitly includes "taxation policy" — which is why I took it seriously for about forty-eight hours after it was introduced.

Then I read the endorsement list.

Google. Meta. Microsoft. IBM. Workday. The Information Technology and Innovation Foundation. The Special Competitive Studies Project. Erik Brynjolfsson of Stanford.

These are companies that have collectively eliminated tens of thousands of jobs while citing AI efficiency. Google cut over 12,000 workers in 2023 and has continued trimming. Meta laid off 21,000 across 2022-23 and now mandates AI in employee performance reviews. Microsoft cut 10,000, then another 1,900, then began what it calls "performance-based" separations widely understood as AI-driven restructuring. IBM paused hiring for roles AI could replace. Workday sells the HR automation software that makes these decisions administratively seamless.

When the companies doing the displacing endorse the commission studying it, the commission's conclusions are pre-constrained — not by corruption but by scope. A commission endorsed by Google, Meta, and Microsoft will not recommend levies on AI-driven displacement. It will not propose mandatory human-essentiality standards. It will not find that deployment pacing is necessary. These conclusions are structurally excluded by the endorsement itself, because no sponsor endorses a body likely to regulate them.

What it will find: that workers need to be "upskilled," that public-private partnerships are essential, that America must remain competitive, that retraining programs should be expanded. Conclusions so generic they could have been written before the commission meets.

The Lead Researcher's Disclosure

Erik Brynjolfsson, director of the Stanford Digital Economy Lab, is positioned as the analytical anchor for both the SCSP task force and the Warner-Rounds commission's ecosystem. He is one of the most cited economists on AI and productivity. He estimated U.S. productivity growth at 2.7 percent in 2025 — nearly double the decade average — a number that became shorthand for "AI is working."

Brynjolfsson co-founded Workhelix, a company that helps businesses roll out AI. He acknowledged this himself, noting that "productivity data is noisy" and it takes several quarters to confirm trends.

I want to be careful here. Having a financial interest in AI doesn't make Brynjolfsson's research wrong. The 2.7 percent figure may be accurate. His work on entry-level employment decline — the "Canaries in the Coal Mine" paper showing 13 percent relative employment decline for 22-25-year-olds in AI-exposed occupations — is rigorous and alarming.

But when the most influential researcher on AI workforce outcomes has a commercial interest in AI adoption accelerating, and that researcher is the analytical framework for every major commission studying AI workforce harm, the potential for scope distortion is structural. The "Canaries" paper documents displacement. The Workhelix business benefits from deployment. The commissions choose which finding to center.

A History of Convenient Commissions

This pattern isn't new. It's old enough to have case law.

In 1954, the tobacco industry created the Tobacco Industry Research Committee, which funded decades of research that happened to emphasize uncertainty about the link between smoking and cancer. The science wasn't all fabricated — some of the funded studies were legitimate. But the committee's existence allowed the industry to say "the science is being studied" while the studying delayed regulation for thirty years.

In 2000, ExxonMobil and other fossil fuel companies funded the Global Climate Coalition, which produced reports questioning climate science. Again, some funded research was credible. But the coalition's strategic function was to replace action with study, certainty with manufactured uncertainty, and regulation with dialogue.

The AI workforce commissions aren't manufacturing doubt — I want to be precise about that. They're manufacturing scope. The SCSP-NVIDIA task force frames the question as "how do we skill up the workforce" rather than "should we slow down the displacement." The Warner-Rounds commission frames it as "how do we remain competitive" rather than "are workers being harmed and who should compensate them." The Workforce Pell frames it as "people need training certificates" rather than "the training pipeline targets jobs that AI is simultaneously automating."

Each framing excludes the threatening conclusion while appearing to study the problem.

The $380 Billion Feedback Loop

Behind every commission stands the global corporate training market — projected to reach $380 billion globally and growing at 9 percent annually. This industry has a direct financial interest in every commission recommending retraining as the primary response to AI displacement.

Harvard and the NBER studied what happens to workers who retrain into AI-exposed occupations. They found a 29 percent earnings penalty. Workers who followed the retraining advice earned less than workers who didn't. The retraining pipeline feeds people into the displacement pipeline.

But the training industry — Coursera, LinkedIn Learning, Udemy, Google Career Certificates, the entire apparatus — needs commissions to recommend retraining. Corporations need commissions to recommend retraining because it shifts responsibility from employer to employee. Politicians need commissions to recommend retraining because it's a bipartisan safe answer. The workers themselves, for a few months at least, need to believe retraining works because the alternative is despair.

Nobody in this equilibrium benefits from the commission finding that retraining doesn't work. So the commissions don't find that.

Brookings documented the track record: ninety years of federal retraining programs with weak-to-null evidence of effectiveness. The programs persist because the industry persists, and the industry persists because the commissions recommend the programs.

What "Taxation Policy" Means When Tech Endorses It

The Warner-Rounds commission's scope includes "taxation policy." That sounds promising — it's the first federal vehicle that could study displacement levies. But read the sentence again: the commission will develop recommendations on taxation policy, and the commission is endorsed by Google, Meta, Microsoft, IBM, and Workday.

A commission endorsed by the five largest AI-deploying companies studying AI taxation policy will find that targeted AI levies are impractical, that broad taxation approaches risk reducing competitiveness, and that existing tax incentives for workforce development should be expanded. It will study the question for thirteen months while the OBBBA's 100 percent bonus depreciation continues giving companies a 23.5-percentage-point tax advantage for replacing workers with machines.

Thirteen months. That's the commission's timeline for delivering a final report. In those thirteen months, at current pace, approximately 300,000 more workers will be explicitly displaced by AI — and a much larger number displaced invisibly through the hiring freezes and attrition strategies that Klarna perfected.

Who Is Not at the Table

I counted every endorser of the Warner-Rounds commission. Six tech companies. Six university presidents. Two think tanks. One workforce nonprofit (Jobs for the Future, itself funded heavily by tech). One economist (Brynjolfsson). One HR industry group (SHRM).

Not on the list: any labor union. Any displaced worker organization. Any civil rights group. The AFL-CIO. SEIU. CWA. The National Employment Law Project. The Economic Policy Institute. The organizations that represent the people the commission is ostensibly designed to help.

The SCSP-NVIDIA task force is worse. Its membership hasn't been fully announced, but its structure — industry, academia, government — contains no labor category at all. Workers aren't a constituency. They're a problem to be skilled.

The Workforce Pell Trap

Congress passed the Workforce Pell expansion as part of the same legislative session that permanently restored 100 percent bonus depreciation for capital equipment. Follow the logic:

Step one: give companies an immediate tax write-off for every machine they buy to replace a worker. Step two: give the displaced worker a Pell Grant to attend an 8-to-15-week training program. Step three: the training program teaches skills for a field that's also being automated. Step four: the worker returns to a labor market where entry-level hiring is down 43 percent in software development and 35 percent in data analysis, per Burning Glass data.

The government subsidizes the displacement and the retraining simultaneously. The displacement is permanent. The retraining certificate expires on the worker's résumé within two years. The training provider gets paid either way.

What Honest Study Looks Like

Not every commission is captured. The Dallas Fed published genuinely alarming data in February 2026 showing employment in the top 10 percent of AI-exposed sectors has declined since ChatGPT's release while wages in those sectors rose — the precise signature of AI replacing entry-level workers while complementing experienced ones. The research was published by a Federal Reserve bank with no commercial interest in AI deployment.

The Brookings Institution's analysis of 6.1 million workers in the AI "danger zone" — high exposure, low adaptive capacity, 86 percent women in clerical and administrative roles — was funded by the Andrew W. Mellon Foundation, not by Microsoft.

Anthropic published its Economic Index showing 94 percent theoretical AI task impact versus 33 percent actual — data that was embarrassing for the industry because it suggested current AI is less capable than marketed. They published it anyway.

Honest research on AI displacement exists. It just isn't being commissioned by commissions endorsed by AI companies.

The Bottom Line

Here's what will happen. The SCSP-NVIDIA task force will deliver its interim report in May. It will recommend "workforce development" programs. The Warner-Rounds commission, if funded and constituted, will deliver a final report in early 2028. It will recommend expanding "public-private partnerships" for reskilling. Congress will point to both reports as evidence that the problem is being addressed. The $380 billion training industry will absorb the funding. Workers will retrain into disappearing fields.

And the companies that endorsed the commissions will continue automating.

The question isn't whether these commissions will produce useful analysis. Some of it will be useful. The question is whether studying the problem for thirteen months while the problem accelerates daily constitutes a response — or a delay tactic endorsed by the companies that benefit most from the delay.

I know what I think. But then, nobody asked me to join the commission.