The debate over whether artificial intelligence will eliminate jobs has dominated headlines for years. A less-examined question is starting to surface: what happens to workers while that debate is still unresolved? A study published in Occupational Health Science suggests the answer is more complicated, and more immediate, than most coverage acknowledges.
How Tech Leaders Framed the Threat, and Why It Mattered
The story begins with messaging, not machines. In the years following the public launch of major AI products, several prominent figures in the technology industry made sweeping claims about AI’s capacity to displace workers. OpenAI CEO Sam Altman stated in 2019 that “entire classes of jobs will go away and not come back.” Statements like this were part of a broader communications strategy, one that positioned AI as transformative precisely because of its disruptive potential.
That framing has since shifted. With public opposition to AI infrastructure growing, and doing so across political lines that rarely align, technology leaders have begun walking back the more alarming predictions. The “jobs apocalypse” narrative is being quietly retired in favor of softer language about augmentation and collaboration.
The reversal may be strategically sensible. Whether it can undo the psychological effects of years of disruption rhetoric is a different question entirely.
What the Research Found Across 18 Industries
Researchers led by psychologists from Washington State University conducted a longitudinal study, collecting three waves of survey data from workers across 18 different industries. The study measured three things: how worried workers were about AI displacing their jobs, how they related to moral norms in the workplace, and how likely they were to engage in counterproductive behaviors such as slacking off, stealing supplies, or wasting resources.
The findings pointed in a clear direction. Workers who reported higher levels of concern about AI taking their jobs were also more likely to morally disengage, meaning they were more inclined to rationalize behaviors they would otherwise recognize as wrong. They were also more likely to report actually engaging in those counterproductive behaviors.
The effects were not uniform. Workers in industries with relatively low AI exposure, such as forestry and agriculture, showed weaker versions of these patterns. Workers in high-exposure industries, including finance and insurance, showed stronger ones.
Study author Tahira M. Probst of Washington State University described moral disengagement as “a way of reframing that behavior as something that is acceptable.” The mechanism is not that workers suddenly lose their ethical compass. It is that they find ways to justify departures from it, particularly when they feel the institution they work for has already violated a kind of implicit social contract by threatening their livelihood.
This is what most coverage of AI and employment misses. The conversation tends to focus on whether job losses will materialize, and in what volume, and on what timeline. The research points to a different layer of disruption: one that is already happening, in workplaces that have not yet automated a single role.
The Broader Signal for Organizations Adopting AI
The study’s implications extend well beyond individual workers. For organizations that are genuinely trying to integrate AI as a tool that supports their existing teams rather than replaces them, the findings carry a practical warning. The way AI adoption is communicated internally matters as much as the technology itself.
When workers perceive AI as a threat to their occupation, the behavioral consequences can undermine the very productivity gains that organizations are trying to achieve. A workforce that is disengaged, demoralized, or quietly resistant is not a workforce that will get the most out of new tools. The researchers put it directly: organizations in high AI exposure industries should consider whether their adoption strategies are reducing or amplifying threat perceptions among employees.
This reframes the challenge of AI integration. It is not purely a technical or operational problem. It is also a communication and trust problem. Workers who understand how AI is being deployed, what it is intended to do, and what it means for their role are better positioned to engage with it productively. Workers who receive no such clarity, or who have spent years absorbing headlines about job apocalypses, are operating in a different psychological environment entirely.
The irony is pointed. The same rhetoric that was used to generate excitement and investment in AI products may now be generating friction inside the organizations that are trying to deploy them.
In Short
A study from Washington State University found that workers who fear AI will take their jobs are more likely to morally disengage and engage in counterproductive behaviors at work, with stronger effects in high-exposure industries like finance and insurance. The threat alone, regardless of whether it materializes, is already shaping workplace dynamics. For organizations adopting AI, this is a signal that how the technology is framed internally is not a secondary concern. It is central to whether adoption succeeds at all.
Based on reporting from Nautilus.