They’re not picketing. They’re not writing manifestos. They’re feeding bad data into training sets, quietly reverting to manual processes, and slow-walking implementation timelines. Across industries from finance to logistics to media, a growing number of Gen Z employees are engaging in what amounts to workplace sabotage — directed squarely at artificial intelligence systems their employers are racing to deploy.
The phenomenon, first reported in detail by Fortune, has caught corporate leadership off guard. Internal surveys and interviews with managers at more than a dozen major firms paint a picture of coordinated, low-level resistance that’s far more sophisticated than simple Luddism. These aren’t workers smashing looms. They’re workers who understand the technology intimately — and who have decided, for a range of reasons, that its unchecked adoption threatens their livelihoods, their professional development, and in some cases, the quality of the work itself.
The scale is difficult to pin down precisely, and that’s partly the point. Acts of AI resistance are designed to look like ordinary friction — the kind of delays and hiccups that accompany any major technology rollout. A corrupted dataset here. A missed integration deadline there. A training session where participation is technically present but substantively absent. According to the Fortune report, HR departments at several Fortune 500 companies have begun investigating patterns of what they’re internally calling “passive non-compliance” with AI adoption mandates.
One mid-level manager at a financial services firm told Fortune that junior analysts had been caught deliberately introducing errors into AI-assisted models, then pointing to those errors as evidence the technology wasn’t ready. “They weren’t wrong that the outputs were bad,” the manager said. “They just made sure they were bad.”
That anecdote captures the essential cunning of the resistance. It exploits the fact that AI systems are genuinely imperfect, that hallucinations and errors are real and documented problems. By amplifying those flaws — or creating conditions where they’re more likely to surface — workers can undermine confidence in AI tools without ever openly opposing them.
So why Gen Z specifically?
The generational dimension matters, though not for the reasons typically cited in trend pieces about young workers. This isn’t about attention spans or entitlement. Gen Z employees, broadly defined as those born between 1997 and 2012, are entering the workforce at the precise moment when AI threatens to collapse the traditional career ladder. Entry-level knowledge work — the research, the drafting, the data processing that has historically served as on-the-job training — is exactly the category of labor most immediately displaced by large language models and automated analytics tools.
The math is bleak for a 24-year-old associate at a consulting firm. If AI handles the grunt work that was supposed to teach her the business, what does she do for three years? And if she doesn’t learn the business from the ground up, how does she advance? The implicit bargain of early-career drudgery — you do the tedious work now, you gain expertise and move up later — breaks down completely when the tedious work is automated away before you’ve had a chance to learn from it.
This isn’t a hypothetical concern. McKinsey’s own research has estimated that activities making up roughly 60% of current occupations could be automated with existing technology. A January 2026 report from the Brookings Institution found that workers under 30 were disproportionately concentrated in roles with the highest automation exposure. And a Deloitte survey published in March found that 67% of Gen Z respondents believed AI would make it harder, not easier, to build a meaningful career.
The resistance takes many forms. Fortune’s reporting identified several distinct patterns across industries. In tech companies, junior engineers have been found to quietly deprioritize AI integration tickets in favor of other work, pushing deployment timelines back by weeks or months. In media organizations, young editors and writers have organized informal pacts to avoid using AI drafting tools, even when explicitly encouraged to do so by management. In logistics and supply chain firms, warehouse-adjacent workers have filed exaggerated bug reports about AI-driven inventory systems, creating a paper trail of dysfunction.
Not all of it is coordinated. Much of it is individual, instinctive, and driven by a straightforward calculation: if I make this tool work well, I’m building my own replacement.
That fear is not irrational. In February, IBM announced it would not replace roughly 7,800 back-office positions that could be handled by AI, a decision CEO Arvind Krishna framed as natural attrition but which sent a clear signal. Klarna, the Swedish fintech company, said its AI assistant was doing the work of 700 full-time customer service agents. And in March, Dropbox laid off 16% of its workforce, with CEO Drew Houston citing the need to “make room for AI.”
But the resistance isn’t purely about job preservation. Conversations with young workers — on platforms like Reddit, in Discord servers dedicated to workplace organizing, and in interviews conducted for this article — reveal a more textured set of motivations. Many express genuine concern about the quality and ethics of AI-generated output. They worry about bias in training data. They worry about the erosion of craft. They worry about a future where mediocre machine output becomes the default standard because it’s cheaper and faster.
“I didn’t go to school for four years to proofread a chatbot,” one 26-year-old copywriter at a New York advertising agency said in a conversation on X. The sentiment, in various forms, echoes across industries.
Corporate leadership is responding with a mix of carrots and sticks. Some companies have launched internal AI “champions” programs designed to identify and reward early adopters, hoping peer influence will overcome resistance. Others have taken a harder line. According to Fortune, at least two major consulting firms have begun incorporating AI tool adoption metrics into performance reviews — effectively making resistance a fireable offense.
Neither approach addresses the underlying anxiety.
The tension playing out in cubicles and Slack channels mirrors a broader societal reckoning with the speed of AI adoption. Public opinion polling from Pew Research Center, published in late 2025, found that Americans were roughly evenly split on whether AI would create more jobs than it destroys. But among adults under 30, pessimism dominated: 58% said they expected AI to have a mostly negative effect on employment.
Labor economists are watching the workplace resistance phenomenon closely. David Autor, the MIT economist whose research on technology and labor markets is widely cited, noted in a recent lecture that the current moment bears some resemblance to earlier waves of automation anxiety — but with a critical difference. Previous automation waves primarily affected manual and routine cognitive labor. This one targets the analytical, creative, and communicative tasks that define professional white-collar work. The people being displaced aren’t just workers. They’re the people who were told they were the future.
And they have tools of their own. Gen Z’s digital fluency — the very quality that makes them attractive hires in a tech-forward economy — also makes them exceptionally effective at quiet resistance. They know how algorithms work. They know where the vulnerabilities are. They understand that an AI system is only as good as the data it’s trained on and the humans who implement it.
Some observers see a silver lining. Ethan Mollick, a Wharton professor who has written extensively about AI in the workplace, has argued that worker resistance can serve as a useful corrective to overly aggressive deployment. “Companies that listen to the friction are going to build better systems,” Mollick wrote in a recent Substack post. “Companies that bulldoze through it are going to get exactly what they deserve.”
That’s a generous reading. A less generous one is that the resistance, however understandable, is ultimately futile — that it delays but doesn’t prevent the adoption of tools that will fundamentally reshape the labor market regardless of what any individual worker does. The history of technology adoption suggests that resistance slows diffusion but rarely stops it. And the competitive pressures driving AI investment are enormous. Companies that don’t adopt will lose to companies that do.
But history also suggests that the terms of adoption matter enormously. The workers who resisted the introduction of automated looms in 19th-century England didn’t stop industrialization. They did, however, contribute to a political and social response — factory regulations, labor protections, eventually the welfare state — that shaped how industrialization unfolded. The question isn’t whether AI will be adopted. It’s whether the people most affected by its adoption will have any say in how.
Right now, the answer appears to be: only through sabotage.
That should concern everyone — not just the workers pulling the strings, but the executives who’ve created conditions where sabotage feels like the only available form of agency. When your youngest, most digitally literate employees are actively working against your technology strategy, the problem isn’t generational attitude. It’s strategic failure.
The companies that figure this out — that find ways to give early-career workers a genuine stake in AI-augmented workflows rather than simply announcing their obsolescence — will have a significant advantage. Not just in retention, but in the quality of their AI implementations. Because the dirty secret of enterprise AI is that it still requires enormous amounts of human judgment to work well. The workers being asked to train, test, and validate these systems are the same workers who have every incentive to make sure they fail.
That’s not a technology problem. It’s a management problem. And unlike a hallucinating chatbot, it won’t be solved by throwing more compute at it.
WebProNews is an iEntry Publication