The Most Concerning 27 Minutes You'll Hear Today: Why AI Labor Disruption Echoes the Luddites

In 1811, the British government sent 12,000 soldiers after its own people. Not invaders. Weavers and knitters. The most skilled workers in England. Their crime was smashing the looms that were replacing them. Seventeen were hanged. Dozens were shipped to penal colonies. Breaking a loom became punishable by death.

Fast-forward to early 2026. Two U.S. lawmakers drop a bill that hits the same nerve: the AI Data Center Moratorium Act, introduced March 25 by Bernie Sanders and co-sponsored in the House by Alexandria Ocasio-Cortez. Their claim is blunt. AI data centers are accelerating a wave of displacement that could wipe out 100 million jobs. When that bill landed, the pattern was impossible to miss. Politicians trying to stop a technology by banning the machines is not new. It started 215 years ago in Nottinghamshire with a man who may not have existed: Ned Ludd.

The insult hides the real story

Call someone a Luddite today and you mean technophobe. Someone who hates their phone. Someone who refuses self-checkout. That is not who the original Luddites were.

They were stocking-frame knitters, croppers, shearers. People who spent years, sometimes a decade, mastering a craft. The elite of the English working class. Around 1811, factory owners in Nottinghamshire installed wide-frame looms that unskilled workers, sometimes children, could run. Quality was worse at first. Speed and price were better. Employers used the machines to cut wages by 50% or more and swap trained craftsmen for whoever would work cheaper.

The skilled workers petitioned Parliament. They negotiated with employers. They tried the courts. Nothing worked. The Combination Acts of 1799 and 1800 had made trade unions illegal. No organize. No bargain. No strike. Every legitimate door was shut. So they smashed machines under the name of Ned Ludd, sometimes Captain Ludd or King Ludd, with letters supposedly from Sherwood Forest. Nobody knows if Ned was real. The legend points to a young apprentice near Leicester who wrecked two frames in a rage around 1779. Thirty years later that story became a mask. Historians call what followed collective bargaining by riot.

They were not anti-technology. They targeted what they called obnoxious machines, the ones used to bypass apprenticeship and quality standards. A Luddite might run a stocking frame by day and smash a different one at night. They hated what certain employers were doing with the tools.

Then the state answered with force. Twelve thousand troops against English workers in English towns. Special trials in January 1813. At least 17 hanged. Dozens deported to Australia for life. The Frame Breaking Act of 1812 made machine-breaking a capital offense. Lord Byron used his maiden speech in the House of Lords to argue against the death penalty. Parliament ignored him.

Right about the problem, wrong about the solution

A few decades later, the machines had won. England became the wealthiest nation on Earth. Real incomes roughly doubled by 1860. Textiles the Luddites tried to protect became the engine of British power. That does not make the short-term pain fictional. Wages did collapse. Families were destroyed. The transition lasted decades of real misery. They saw the wound clearly. They picked the wrong cure. That distinction is the whole story.

What the 2026 bill actually does

The public story around the Sanders-AOC bill is mostly right: AI is getting powerful, jobs are at risk, data centers gulp energy, something must be done. Sanders' October 2025 office report claimed roughly 100 million U.S. jobs are at risk, about two out of three. He has floated a robot tax on firms that replace people with automation, a 32-hour workweek with no pay cut, 20% of stock in large corporations held by workers, and 45% of board seats for employees. AOC called the moratorium a moral obligation, tied it to rising local electricity bills, and pledged politicians should refuse AI company money.

Here is the hard part. The bill does not regulate how AI is used. It does not create a fund for displaced workers. It does not build retraining. It does not invest in energy to offset data-center load. It bans buildings: the physical data centers that run the models. That is trying to stop cars by banning roads. You are not hitting the technology. You are hitting the infrastructure it needs inside your borders. The technology can leave.

Sanders' broader ideas, the robot tax, shorter week, worker ownership, are worth arguing about because they engage the real problem: how gains get shared. The moratorium is different. It tries to stop the thing itself. History has a clear verdict on that move.

Same fear, same tactic, same outcome

In 1865 Britain passed the Red Flag Act. Any self-propelled vehicle on a public road needed a crew of three, plus a person walking 60 yards ahead with a red flag by day or a lantern by night. Speed limits: 4 mph in the country, 2 mph in towns. You walk faster than that. Railway companies, horse breeders, and carriage interests lobbied for it. Full relief only came in 1896. Thirty-one years of choke regulation. By then France held nearly half the global auto market. Germany had Benz and Daimler. America was heading toward Ford. Britain, which invented the steam engine, got left behind in cars for a generation. The Red Flag Act did not save the horses. It guaranteed the auto boom happened somewhere else.

Early 1900s America: farmers hated cars. Devil wagons. They scared horses, tore dirt roads, raised dust that ruined crops. Some towns forced drivers to stop, kill the engine, and wait for the animal to pass. Within 25 years those same farmers bought Model Ts built for rural roads. Tractors followed. One farmer who once fed four people could feed 40. The hardest opponents became the most dependent adopters.

1979: Three Mile Island. Anti-nuclear politics froze new U.S. plants for decades. More than 100 planned reactors canceled. Now, in 2026, the country is scrambling for nuclear again because AI data centers need huge clean power. Microsoft signed to restart Three Mile Island itself. Google is chasing small modular reactors. Amazon is buying nuclear-powered sites. The thing America banned for forty years is the thing it desperately needs.

1990s Europe: GMO crops. The EU banned or heavily restricted them while the U.S. embraced them. American yields jumped. EU farmers fell behind. Europe ended up importing the soybeans it had refused to grow. Food quality stayed high. Food security got worse.

Four technologies. Same loop. New tech threatens jobs and power. Legitimate fear drives restrictions. Restrictions do not kill the tech. They move it. The banning country falls behind. The workers the ban was supposed to protect end up worse off.

The risks are real. The pause is fake.

Mocking Sanders and AOC as modern Luddites is the easy move. It is also incomplete. Energy costs are real. A large AI data center can draw about 100 MW, roughly 80,000 homes. Some residents have watched bills jump around 30%. Wealth concentration is real too. Gains flow to people who already own capital and can capture the labor savings. The speed may outrun anything prior. Luddites faced mechanical looms that did one job. Today systems write code, draft legal briefs, read medical images, design buildings, translate in real time, and hold conversations that pass casual Turing tests.

MIT economists Daron Acemoglu and Pascual Restrepo have shown that since about 1987, automation displacement has outpaced reinstatement. From 1947 to 1987 the two were roughly balanced. After 1987, displacement rose toward 0.7% a year and reinstatement fell toward 0.35%. Over three decades that gap compounds into roughly a 10% net cut in labor demand. It helps explain stagnant wages for lower-skilled workers even as the economy grew. Robots did take jobs. The open question is what we do next.

Exposure is not replacement. A radiologist is exposed because AI can read certain images. That does not mean radiologists vanish. It means radiologists who use AI replace those who refuse. Conflate those two and you can inflate a scary 100 million.

Ban the buildings and the models leave

A U.S. moratorium on new AI data centers will not stop AI. Not by a day. It will move training to Shenzhen, Abu Dhabi, Singapore, or wherever the permits and power are. China is pouring hundreds of billions into AI through 2030. The UAE is building desert campuses. Singapore is racing with tax breaks and fast permits. Those countries are not debating whether to build. They are racing.

Carl Benedikt Frey's Technology Trap argument still holds: transitions produce huge long-run gains and real short-run crises if the displaced are left with nothing. His conclusion is not ban the tech. Manage the transition.

What the Luddites actually left behind

The Luddites did not just vanish after the hangings. Machine-breaking was the only tool left when unions were illegal. The unrest helped force Parliament to repeal the Combination Acts in 1824. Unions became legal. Then came the Factory Acts: bans on young child labor, hour caps, inspectors with real power, safety rules. The seed of smashed looms became the tree of worker protections we now treat as normal.

If that is the pattern, the point is not whether the Sanders-AOC bill is perfect. It will not pass. Too many states want the data-center jobs and tax base. Too many sites already have dirt turned. The noise might still matter. Pressure can produce a retraining fund, energy offsets, or a transition tax even when a freeze on buildings dies. Stopping the machines has never worked. Fighting for workers to share the gains has. That is the fight worth having.

The most concerning 27 minutes are not a prophecy that AI ends work. They are a warning that we keep reaching for the same failed switch: shut off the infrastructure and hope the disruption stays home. It will not. It will leave. And the bill for that exit lands on the people the ban claimed to protect.

The workers who felt the first cut of the industrial loom were not crazy for noticing. They were wrong for thinking the smash would stick. The same is true for AI. Notice the displacement. Fight over who captures the surplus. Do not confuse that with a freeze on the racks that train the models. One of those fights can still end with a broader middle class. The other ends with the same models running in another country while your power bills and job base both lose.

Check the video here.