Nearly a third of workers sabotage company AI—wage cuts, not layoffs, are the real threat

In mid-July 2026, Apollo Global Management's chief economist Torsten Slok published a paper with co-author Sania Edlich that challenged his own earlier optimistic predictions about AI and employment. Just two months prior, on May 29, Slok had published a Daily Spark titled "Zero Evidence of AI-Related Job Losses," invoking the Jevons Paradox to argue that efficiency gains would expand demand and create net job growth. But when Slok and Edlich analyzed actual usage data from Anthropic's Economic Index—real Claude interaction logs showing what workers were genuinely doing with AI—they found something different: wage compression without significant job loss.

Their analysis spanned 321 occupations matched to Bureau of Labor Statistics data from 2015 to 2025. The key finding: workers in high-AI-exposure occupations experienced real wage growth that slowed by 6.7 percentage points relative to less-exposed workers after 2023, with no statistically significant employment effect. The pain was heavily concentrated at the bottom of the income ladder. Workers in the bottom wage quartile saw wages down 10.7% relative to low-exposure occupations; the second quartile down 5.4%; the third quartile down 4.0%; while the top quartile experienced no significant effect. Service occupations took the steepest hit at down 24.3%, though the authors cautioned this figure was based on a small subsample. Overall, approximately 5.8 million U.S. workers—about 3.7% of the labor force—sit in occupations exposed enough to feel this squeeze, amounting to a conservative $28 billion in annual labor income loss that the authors expect to keep climbing.

The finding arrived amid widespread worker anxiety and resistance. A June 2026 survey by Software Finder of 1,005 employed U.S. workers found that half described themselves as actively resisting new AI tools. Workers who resisted AI earned roughly 20% less on average than those who embraced it ($65,645 versus $81,526), though 45% cited fear of becoming replaceable as their reason for holding back, and only 16% believed their company was adopting AI for genuine business value rather than hype or competitive pressure. Resistance sometimes escalated into deliberate sabotage: an April 2026 survey of 2,400 knowledge workers across the U.S., U.K., and Europe by Writer and Workplace Intelligence found that 29% of employees admitted to actively sabotaging their company's AI strategy—a figure that jumped to 44% among Gen Z workers. The sabotage took concrete forms: entering proprietary company information into unapproved public AI tools, using unauthorized "shadow AI" systems, refusing to engage with company-mandated tools, tampering with performance reviews, and deliberately producing low-quality work to make AI look ineffective. Of the workers who admitted to sabotage, 30% cited fear that AI would take their job as their primary motivation. Additional findings showed that 13% of workers admitted to faking AI use—appearing to use a tool while doing the task manually—and only 6% believed their managers accurately understood how often employees actually used the tools their companies had rolled out.

Slok's paper contradicted analysis from Anthropic's own head of economics, who in late July published a lengthy essay on X drawing on 18 months of internal research. He concluded that the U.S. labor market had "not yet taken a visible hit from AI," pointing to a 4.2% unemployment rate (which the Federal Reserve considers full employment), with job openings roughly matching the number of unemployed workers and prime-age employment near multi-decade highs. Yet the two analyses need not contradict—they answer different questions with overlapping data. It is entirely possible for a labor market to show flat unemployment while wages quietly fall in relative terms. A comprehensive literature review cited by Reuters in July found that most datasets showed "little evidence of economy-wide job loss or wage decline," attributing AI's impact so far to "task reallocation and within-firm productivity gains, rather than mass displacement." This conclusion sits uneasily next to Slok's wage-compression findings, explaining why reasonable economists looking at adjacent data reach opposite-sounding conclusions.

Slok acknowledged his paper's limits—the exposure measure relies solely on Anthropic's data, and only 321 of roughly 800 BLS occupations could be matched—yet remained unambiguous about the stakes. He wrote: "The critical policy question is not whether AI will reshape the labor market more broadly, but how quickly, and whether workers will have the support they need when it does." The wage-compression pattern echoes historical technological disruption. Textile mechanization crushed wages for hand-loom weavers well before it created higher-paying factory jobs elsewhere, giving rise to the Luddite movement. More than a century later, industrial robotics in manufacturing during the 1980s and '90s coincided with decades of stagnant real wages for blue-collar workers even as productivity climbed steadily, originating the "Rust Belt." Financial Times analyst Joel Suss found that gains from new technology have not automatically flowed to the workers producing them since around 1970, as labor's share of GDP has fallen relative to capital's across the U.S., Japan, and most of Europe. "Insofar as advances in AI constitute capital-biased technological change," he argued, "the pay-productivity gulf will widen further." What emerges is a labor story that resists the clean narrative either side wants to tell: not the mass-layoffs scenario Anthropic CEO Dario Amodei has warned about, nor the all-clear that unemployment data suggests, but something quieter and more corrosive—a mechanism that shows up in paychecks rather than pink slips, indistinct enough that reasonable economists can reach opposite-sounding conclusions.