Sixty percent of workers already worried about layoffs at their company have used AI to take over a task that used to belong to a coworker.
At companies that already ran layoffs in the past year, that number climbs to seventy-four percent, among that same layoff-anxious group.
Seventy percent of the time, the coworker whose task got absorbed was a workplace friend. Fifty-seven percent of those friends were later laid off. Sixty-three percent of the people who did the absorbing never told their manager AI was doing the work.
Read that sequence slowly, because it inverts the story most companies tell themselves about why employees go quiet about AI.
The Story Everyone Assumes Is The Wrong One
The assumed story is fear. An employee worries AI makes them replaceable, so they hide their usage to avoid drawing attention to how fast they've gotten. That story is real, and it shows up in the data. But it isn't the dominant one.
Ask employees directly why they keep AI use secret, and the top answer isn't fear of losing their own job. It's wanting a "secret advantage" over their peers, named by thirty-six percent of workers, ahead of job-loss fear at thirty percent. That's not a rounding difference. It means the plurality of secret AI use is offense, not defense.
The ResumeBuilder numbers make the offense concrete rather than abstract. Among workers already anxious about layoffs, a majority have actively used AI to absorb a colleague's responsibilities, and at companies already cutting staff, that number rises past seventy percent.
Nearly a third of the time, they take on four or more of a colleague's responsibilities within six months. This isn't defensive concealment. It's a quiet, individual competition for whose job survives the next round, run through a tool nobody else in the room knows is in play.
Why This Changes What Leadership Actually Has To Fix
A fear-based problem has a fear-based fix. Say the reassuring words, mean them, and a workforce responds by trusting you a little more each time the promise holds. That's the fix most companies reach for, and it's the wrong tool for a competitive dynamic instead of an anxious one.
An employee absorbing a colleague's tasks for competitive advantage isn't waiting for permission or reassurance. They're calculating that being the person who can visibly do more, quietly, beats being the person other employees can see struggling. No leadership speech changes that calculation, because the speech was never what was driving the behavior. The math was.
This is why the data on friends is the part that should actually worry a leadership team. Seventy percent of the time, the task got absorbed from someone the employee considered a friend. Fifty-seven percent of those friends were later laid off. That's not a story about strangers competing in an abstract labor market. It's colleagues who know each other, one quietly building a case for their own indispensability using the other's work as raw material, without ever naming what's happening out loud.
There's a specific reason this dynamic is so hard for a company to see happening in real time.
Nothing about it looks like sabotage from the outside. The employee absorbing more tasks looks like exactly what most performance reviews reward: someone doing more, contributing across a wider range of work, becoming harder to lose. The colleague quietly losing ground looks like exactly what most performance reviews already suspect: someone falling behind, contributing less, becoming easier to justify cutting. The tool doing the actual work never shows up on either side of that story, which means the entire dynamic reads as a normal, even healthy, performance gap opening up between two employees, right up until one of them is let go and the other quietly picks up whatever's left.
What Actually Has A Chance Of Working
Deloitte's own research on this points toward a different lever than reassurance. Two-thirds of UK workers have already used generative AI at work, and sixty-three percent worry their manager might conclude the tool can do their job. The researcher who ran that survey, Paul Lee, put the actual tension in one sentence: workers are thinking this could increase my profitability or this could diminish my worth, at the same time. That's not confusion. It's an accurate read of a genuinely unresolved question nobody in leadership has actually answered.
The fix that might work is not a promise about job security. It's making visible, deliberately and specifically, what AI fluency is worth to a career inside that company, in terms concrete enough that competing for advantage stops requiring secrecy to work. If getting better at AI is something an employee wants credit for, they show it. If getting better at AI is something they suspect will be quietly punished or used to justify someone's redundancy, they hide it and, apparently, sometimes use it against the person sitting next to them instead.
The organizations that get ahead of this won't be the ones with the most reassuring town hall message. They'll be the ones that made competing for advancement through visible AI skill a better strategy than competing for survival through quiet displacement of a colleague, because for a meaningful share of the workforce, the second strategy is currently winning.
There's a concrete test a leadership team can run to find out which strategy is actually winning inside their own organization, and it doesn't require another survey. Look at the last several rounds of promotions and the last several rounds of layoffs side by side. If the people getting ahead are the ones who talked openly about how AI changed their workflow, competing on visible skill is winning, and the incentive structure is pointed the right way. If the people getting ahead are simply the ones who ended up doing more, with no visible explanation for how, and the people who disappeared were quietly doing less for reasons nobody investigated, the quiet version of this competition is already running, whether or not anyone in the room has named it.
That test costs nothing and requires no new policy. It only requires someone willing to ask an uncomfortable question about patterns that already exist in the company's own personnel records, and to sit with an answer that might not be the one the last town hall promised.
References
ResumeBuilder.com. "3 in 5 Workers Are Secretly Using AI to Take Over a Coworker's Job to Save Their Own." resumebuilder.com, April 23, 2026 (survey of 1,000 full-time U.S. workers concerned about layoffs).
Braun, Sara. "Employees Are Using AI at Work But Hiding It From Their Bosses Because They Think It Gives Them a 'Secret Advantage' Over Their Peers." Fortune, fortune.com, May 29, 2025, citing Ivanti's "Tech at Work" report.
Bloomberg. "UK Workers Hide AI Use From Their Bosses, Deloitte Survey Finds." bloomberg.com, September 16, 2026.
