Asger Risom
New research shows AI isn’t taking jobs—it’s shrinking paychecks. As companies use productivity gains to compress worker wages by up to 10.7%, nearly a third of employees admit to actively sabotaging corporate AI initiatives in retaliation.
For years, the public debate around artificial intelligence in the workplace has focused on a single fear: mass layoffs. But new economic data suggests the technology’s most immediate damage isn’t pink slips—it is smaller paychecks. As employers capture productivity gains without sharing the financial upside, workers are quietly revolting, with nearly a third admitting to active sabotage.
The wage-compression trap
According to a landmark study by Apollo Global Management highlighted in a recent Fortune report, artificial intelligence is acting as a “wage-eating machine.” Rather than relying on theoretical risk models, Apollo chief economist Torsten Slok and co-author Sania Edlich analyzed actual Claude usage logs from Anthropic’s Economic Index across 321 occupations matched to U.S. Bureau of Labor Statistics data.
Their findings reveal that while employment levels in AI-exposed roles remain steady, real wage growth for workers in high-exposure occupations slowed by 6.7 percentage points relative to unexposed roles after 2023. Employers are keeping headcounts flat while using AI efficiency gains to compress wages—assigning workers higher workloads for less relative pay.
This economic pressure hits the lowest-earning workers hardest. For those in the bottom income quartile, real wage growth plummeted by 10.7% relative to low-exposure roles, while workers in the second and third quartiles saw drops of 5.4% and 4.0%, respectively.
Conversely, the top income quartile experienced no statistically significant negative effect, indicating high earners are far better positioned to leverage AI for personal financial gain. All told, approximately 5.8 million U.S. workers—roughly 3.7% of the labor force—are already affected, representing an estimated $28 billion in lost annual labor income.
This wage-compression dynamic bridges the gap between macroeconomic data showing low national unemployment and ground-level worker anxiety. As Financial Times analyst Joel Suss noted, technological productivity gains have historically favored capital over labor since 1970—a trend now accelerating under the enterprise AI rollout.
“Shadow AI,” fake usage, and outright sabotage
Faced with stagnant compensation and expectations to do more work in less time, employees are resisting corporate AI mandates through increasingly aggressive tactics. A June 2026 survey of 1,005 U.S. workers by Software Finder revealed that half actively resist new AI tools, with 13% admitting to “faking” AI usage by pretending to use automated tools while completing tasks manually to appease management.
However, resistance frequently escalates past passive avoidance. An April 2026 survey of 2,400 knowledge workers across the U.S., UK, and Europe conducted by Writer and Workplace Intelligence found that 29% of employees admit to actively sabotaging their company’s AI strategy—a figure that leaps to 44% among Gen Z workers.
The admitted forms of workplace sabotage are highly disruptive. Employees reported leaking proprietary corporate data into unapproved public AI systems and operating unauthorized “shadow AI” tools outside enterprise security perimeters. Others confessed to tampering with performance evaluations, intentionally producing poor-quality work to discredit the AI tools, or flatly refusing to engage with mandatory corporate platforms.
An economic impasse
The research exposes a sharp divide within the tech and economic sectors. While Anthropic’s head of economics has pointed to steady unemployment figures to argue the labor market remains unaffected, Slok’s research demonstrates that flat employment and eroding relative pay can—and do—coexist.
By failing to pass AI-driven productivity gains down to staff, enterprises are creating a self-defeating feedback loop. Workers recognize that adopting automated tools makes them more productive without making them richer, driving an underground resistance movement that threatens to undermine corporate AI investments from the inside.
