In the first half of 2026, 101,743 U.S. layoffs were explicitly attributed to AI — nearly double the total for all of 2025. Tech and finance were shedding an average 28,000 payroll jobs per month, and the July jobs report showed the U.S. economy unexpectedly cut 23,000 jobs overall. In July, nearly 200 economists signed an open letter warning that AI could trigger disruption "larger than the Industrial Revolution." Among them: two Nobel laureates who had previously dismissed displacement fears. And by the end of that month, a survey found that nearly 1 in 3 workers was actively sabotaging their company's AI rollout.

1. Companies Using AI Are Hiring More (Ramp economist Ara Kharazian, OpenAI's Sam Altman, PwC)

Heavy AI investors are growing their teams, not shrinking them — and they have a study of 21,000 firms to back it up.

Heavy AI users are actually hiring more. Ramp, the expense-management company, tracked 21,000 U.S. firms over two years and found that heavy AI adopters grew headcount by 10% — and entry-level hiring by 12%. Ramp's lead economist Ara Kharazian says companies blaming AI for cuts had over-hired during the pandemic. "They're blaming it on AI," he said. "But what we're seeing from firms using AI that didn't have that overhiring problem is that they're continuing to grow."

PwC's data from a billion job ads points the same way. Its 2026 AI Jobs Barometer found that companies in the most AI-exposed sectors grew headcount 52% — versus 36% for less AI-exposed peers. PwC calls these "professionalised" jobs — roles where humans use AI for judgment work. Those roles saw 42% faster salary growth than roles where AI handles the work directly.

The entry-level white-collar apocalypse hasn't materialized. Speaking in Sydney in May 2026, OpenAI CEO Sam Altman said he'd expected more displacement of entry-level white-collar work "by now than has actually happened." Apollo's chief economist Torsten Slok went further in June, writing that there was "zero evidence of job losses because of AI."

2. But Young Workers Are Already Taking the Hit (Stanford's Erik Brynjolfsson, MIT's Daron Acemoglu, ADP's Nela Richardson)

The aggregate numbers look stable. Zoom in on workers aged 22-25 in AI-exposed occupations and the trend only goes one direction.

Aggregate job numbers hide what's happening to young workers. Stanford economist Erik Brynjolfsson and ADP chief economist Nela Richardson built a dashboard tracking 4.6 million workers across 730+ occupations — about 1 in 6 American workers. Their finding: workers aged 22–25 in the most AI-exposed roles are losing employment at 3.8% annually as of April 2026. The rate has grown by roughly half a percentage point every month without reversing. Workers aged 35–40 in those same jobs are growing at 2% annually. "Whatever it is, it's not going away," Brynjolfsson said.

The displacement skeptics reversed course this summer. MIT economists Daron Acemoglu and Simon Johnson shared the 2024 Nobel Prize in Economics and had previously pushed back on AI job-loss fears. Both signed the July 13 "We Must Act Now" letter. It warned that AI "may become radically more powerful over the next 10 years" and could drive a transformation "larger than the Industrial Revolution, but unfolding over a vastly shorter time frame." The letter carried nearly 200 signatories, including 16 Nobel laureates and OpenAI CFO Sarah Friar.

Morgan Stanley's August 5 analysis confirmed the youth-specific pattern. AI has added just 15 basis points to the overall U.S. unemployment rate — but 50 basis points in AI-exposed industries specifically. Workers aged 22–27 are "the age group most affected by AI-related labor market changes" in the first half of 2026.

3. ...And Workers Are Pushing Back on Their Own (29% of U.S. workers, the ILA, the WGA)

Nearly 1 in 3 workers is sabotaging their company's AI rollout. The surveys suggest they already know why their paychecks shrank.

Workers who resist AI are already earning less for it. Apollo Global's July whitepaper found that workers in high-AI-use occupations saw their real wage growth lag low-exposure workers by 6.7 percentage points after 2023. The hit was largest for service workers and bottom-quartile earners. Apollo found no net job losses in aggregate yet — but that wage compression costs an estimated $28 billion in annual labor income across 5.8 million workers.

A third of workers admitted to sabotaging their company's AI strategy. A study of 2,400 knowledge workers by enterprise AI company Writer and research firm Workplace Intelligence found that 29% of employees admitted to actively undermining their company's AI strategy — rising to 44% among Gen Z specifically. The most common form: entering sensitive company data into unauthorized public AI tools. Others produce intentionally bad AI output, fake compliance while completing tasks manually, or simply refuse to engage. Thirty percent said fear of losing their job was the primary motive.

Unions have moved from individual pushback to contract clauses. The International Longshoremen's Association secured a full ban on autonomous port technology. The Writers Guild of America won a prohibition on AI-generated scripts. Las Vegas hotel workers won a right to bargain before any AI decision. Illinois already requires companies to notify workers when AI influences hiring, firing, or disciplinary decisions — a law that took effect January 1, 2026. Colorado's AI Act, which adds a right to appeal AI-driven employment decisions, takes effect January 1, 2027.

Where This Lands

Brynjolfsson says the data is unambiguous for young workers in AI-exposed roles, and the trend has only steepened month over month. Kharazian's counter is that companies that built on AI without a pandemic-era hiring binge are still growing headcount, and the doom narrative mistakes a correction for a collapse. The workers in the middle — losing wages, sabotaging rollouts, and winning union clauses — are acting on what they already know. The protections economists are calling for — severance standards, notification laws, revenue-sharing models — exist on paper but not yet in most workplaces.

Sources