The hidden costs of workplace surveillance

Workplace surveillance is entering a new phase. AI monitoring tools designed for businesses of all sizes are expanding in scope, scale, and influence—raising new questions about autonomy, fairness, and the future of work. In April, Meta—the world’s largest social media company—began deploying a new surveillance tool on the laptops of its U.S.-based workers. Known as the Model Capability Initiative (MCI), the software captured real-time data, including mouse movements, clicks, keystrokes, and screenshots of work applications.

CEO Mark Zuckerberg framed the move as a competitive necessity in the AI race, arguing that employee data could help train the company’s models. The rollout raised immediate concerns, however. It coincided with news that the company would lay off 10% of its workforce in May, intensifying employees’ fears about how their data might be used. Some employees protested that they were being asked to help train the very systems that would replace them. Several organized walkouts and petitions, or they simply delayed installing the MCI software.

The response highlighted a broader reality: In the United States, workers have limited protections against this kind of monitoring. Unlike Meta workers in the European Union—who were exempt from the program because of stricter data-consent requirements—U.S.-based Meta workers have few legal tools to challenge how they are tracked or how their data are used. “The legal question may be whether monitoring is permissible,” said Dennis P. Stolle, JD, PhD, executive lead psychologist for applied psychology at APA. “The psychological and strategic question is whether it is wise, effective, and sustainable. Done wrong, monitoring can contribute to employee attrition and reputational harm for organizations.”

Meta’s approach is part of a larger shift. Workplace monitoring has evolved from punch clocks and supervisor observation into systems where workers operate under constant, often invisible oversight. A 2025 survey by VPN company ExpressVPN found that 78% of U.S. employers use digital tools to monitor employees, including biometric tracking, location data, and AI-driven productivity analysis. Employers can now measure not only what workers do but how they move, communicate, and even appear to feel while doing it.

In practice, the monitoring can be pervasive. Remote employees may have their keystrokes recorded or emails analyzed for tone. In warehouses, sensors flag “time off task,” measuring how long workers pause for breaks or step away from their stations. Many of these systems now go beyond simple tracking: AI-powered “algorithmic managers” can assign tasks, schedule shifts, evaluate performance, and in some cases even initiate terminations. This model, once most visible in gig work, is spreading across retail, health care, logistics, and office environments.

Behind these systems is a rapidly expanding pool of data—often collected without a clear purpose. “Organizations don’t know why they’re collecting the data,” said Tara Behrend, PhD, an industrial and organizational (I/O) psychologist and the John Richard Butler II Endowed Professor at the School of Human Resources and Labor Relations at Michigan State University. “However, they do have a sense that data is powerful, and they should have a lot of it. They’ll figure out what it all means later.”

This tendency sets the stage for what researchers call “function creep,” or data gathered for one legitimate purpose that’s later repurposed without employees’ knowledge or consent. Jessica Vitak, PhD, an information scientist and the director of the Privacy Education and Research Lab (PEARL) at the University of Maryland, cautions that organizations often assume more data automatically leads to better decisions—a false belief that risks reducing workers to an incomplete data portrait. “The reality is that keystrokes, emails, and other captured data are only part of the picture,” she said. “We shouldn’t treat it as if that is the full understanding of what a worker is doing.”

Workplace surveillance is not inherently harmful. In many contexts, it plays an essential role in maintaining safety, ensuring regulatory compliance, and, when implemented well, supporting performance management. “Workplace surveillance is an integral part of the employment contract,” said Kirstie Ball, PhD, a professor of management at the University of St. Andrews in Scotland who studies surveillance in organizations. “It can enable feedback and good management practices.”

Significant problems arise, however, when surveillance becomes “too intrusive, authoritarian, or excessive,” she added. “It can start to upset the psychological contract—those unwritten assumptions that we have when we enter the workplace about how we can expect to be treated by our employer.” Surveillance that violates those expectations can erode trust and strain the employer-employee relationship rather than improve performance.

As digital surveillance becomes more pervasive, researchers are gaining a clearer picture of how it affects workers. APA’s 2024 Work in America survey found that 51% of workers who knew they were monitored reported feeling tense or stressed during a typical workday, compared with 38% of workers who said they were not monitored or did not know whether they were monitored. Workers who knew they were monitored were also more likely to report feeling micromanaged (47% vs. 23%) and planning to look for a new job in the following year (39% vs. 21%). Worry about monitoring was more prevalent among Hispanic, Asian, and Black workers than White workers (46%, 41%, and 37% vs. 29%, respectively).

Behrend describes pervasive monitoring as placing a heavy cognitive and emotional burden on workers that’s akin to forcing a performer to remain permanently on stage. “You’re always thinking a little bit about how you’re being perceived and making sure that you’re putting forward the right image or the right behavior. It’s very taxing. People can’t keep that up all day, every day without burning out,” she explained.

Daniel Ravid, PhD, an I/O psychologist and the director of the Work and Technology Changes (WATCH) Lab at the University of New Mexico, linked that sustained pressure directly to longer-term harm. “Research going back to the 1960s on social facilitation has shown that the mere idea of being observed or being watched does increase stress. When you see high levels of stress that are sustained over long periods of time that are very intense, that’s where you start to see burnout,” he said. “If you’re hitting the level where people are stressed about going to the bathroom, you’ve probably gone too far.” Ravid, Behrend, and colleagues’ meta-analysis of electronic performance monitoring found no evidence it improved worker performance and consistently linked it to increased worker stress (Personnel Psychology, Vol. 76, No. 1, 2023opens in new window).

The stress of surveillance is not evenly distributed. Workers in precarious roles—those with less job security or bargaining power—are more likely to face intensive monitoring and to experience its negative impacts. Drawing on data from APA’s 2023 Work in America Survey, Behrend and colleagues found that workers in precarious positions were more likely to be subjected to electronic performance monitoring, to face more intensive tracking, and to report more negative psychological reactions to surveillance (Journal of Business and Psychology, 2026opens in new window).

A 2025 report by the U.S. Government Accountability Office (GAO) identified a mix of benefits and risks from digital surveillance. Wearables and sensors can improve safety by detecting fatigue, cardiac issues, or hazardous conditions such as extreme heat or toxic exposure. For remote or solo workers, monitoring may also increase a sense of security. However, those same systems can create pressure to perform at unsafe levels. When productivity metrics are tightly tracked, workers may feel compelled to move faster, take fewer breaks, or ignore physical warning signs, raising the risk of injury or exacerbating existing ailments.

Even as organizations collect more data than ever, they often miss essential aspects of work. Surveillance tools consistently fail to capture essential contributions—research, reflection, mentoring, relationship-building—leaving workers vulnerable to evaluations based on an incomplete picture. Researchers describe the result as the creation of “datafied selves,” where workers feel reduced to the metrics they generate rather than recognized for their broader contributions (Pawirosumarto, S., & Kurniawan, B., Journal of Business Ethics, 2026opens in new window). “When we over-rely on that data without taking in contextual factors, without understanding the relationships between individuals in a company, it leads to problems in the decision-making,” Vitak said.

Vitak and Michael Zimmer, PhD, director of the Center for Data Ethics at Marquette University, have identified function creep, in which surveillance tools justified by safety or security morph into performance management tools. In this way, data initially collected for safety or security—such as health monitoring or workplace access—would later be used to evaluate performance or inform promotion decisions (Journal of Computer-Mediated Communication, Vol. 28, No. 4, 2023opens in new window).

Vitak and Zimmer’s research showed that for some groups, the stakes felt even higher. Women, for example, expressed particular concern about the potential reuse of health-related data, fearing it could lead to pregnancy discrimination. “People who might already be vulnerable or feeling less power in the workplace were more concerned about some of these new forms of workplace monitoring. That was often women or people of color who are already subjected to different forms of surveillance,” Zimmer said.

The 2025 GAO report found that some tools used to track employees’ expressions are more likely to wrongly label people from certain racial groups as having negative emotions. Productivity benchmarks may also negatively impact workers with disabilities, such as neurodivergent or other employees who may require more time for cognitive processing between tasks. “What we see is that the most marginalized groups are further experiencing harms from this datafied society,” Vitak said. “The people who aren’t well-off have to give up their privacy more often, and they’re the ones most likely to be harmed by these kinds of programs.”

Surveillance can also create a perverse incentive. Workers learn quickly what is being measured and optimize for exactly that. “What you’re doing when you’re tracking these behaviors is you’re shaping people’s attention and creating a really strong motivation for them to optimize that metric at any cost,” said Behrend. “That can mean taking shortcuts or doing things that are really unsafe or creating really unhelpful competition between people. And all those things end up costing more money over time.”

The results can be counterproductive. Workers may leave files open to maintain the appearance of activity, generate meaningless prompts to keep their mandated AI tools busy, or use mouse jigglers—devices that simulate cursor movement—to fool activity-tracking software.

Reeshad Dalal, PhD, a professor of I/O psychology at George Mason University in Northern Virginia, sees this as an inevitable consequence of measuring the wrong things. “We’re creating an arms race between employers and employees for no purpose on these completely inane metrics that have nothing to do with performance,” he said. And the more tightly workers are controlled, the worse they tend to respond. Psychological science labels this response reactance—the well-documented tendency to push back against perceived threats to autonomy. “People don’t like being told exactly what to do,” Dalal said. “They want some amount of autonomy, and if you force them to do very specific things, they’re going to respond badly.”

Chase Thiel, PhD, an I/O psychologist and professor of management at the University of Wyoming, has studied this dynamic. In research published in the Journal of Management (Vol. 49, No. 2, 2023)opens in new window, he and colleagues found that monitored employees were significantly more likely to cheat, ignore instructions, slow their work pace, and engage in other deviant behaviors because monitoring had eroded their feelings of agency.

“Once you lose that sense of autonomy, you also lose your sense of responsibility,” Thiel said. “It’s not even that there’s an active, conscious gamification of it. It’s more, I’m doing what I’m told. Workers are not invoking their values and principles when making decisions about their own behavior at work. They’re kind of just responding to external controls and stimuli.”

Context matters, however; when monitoring was tied to development rather than punishment, workers trusted their managers more, performed better, and engaged in fewer counterproductive behaviors overall.

According to Cornelius König, PhD, a professor of I/O psychology at Saarland University in Germany who also studies counterproductive work behaviors, what employers say about monitoring systems matters less than whether workers trust their leadership to begin with. For workers, “it really depends on whether you trust your employer,” König said. “If you do, you may be more willing to see the potential benefits of monitoring. But if you don’t, any reassurance they give may be unpersuasive. It doesn’t matter what the official framing is.” The consequence, he argued, is self-reinforcing: Perceived distrust breeds distrust in return. “In the end, if I think this is just a tool to monitor me, and it’s showing distrust in me, I’m not trusting the employer either. And then you get negative effects.”

What should ethical monitoring look like in practice? Thiel offered a three-part test. First, monitoring should have a clear and transparent purpose—workers should know exactly what data are being collected and why. Second, it should have a developmental or safety orientation rather than a punitive one: “You’re not doing this to catch people or because you assume people are trying to take advantage. You’re doing it because you’re trying to find new insights. People are hardwired to want growth. If you tap into that, and that’s the focus of the monitoring, you’re in a good space.” Third, monitoring should be limited in scope. “It can’t be everywhere,” Thiel said. “People need downtime to have an opportunity for creative thought. They’re going to have a hard time experiencing that if you’re monitoring every aspect of their work.” When data are necessary, Thiel recommends collecting them at an aggregated, anonymized level—enough to identify patterns but not to scrutinize individuals. And, most importantly, turn surveillance off when it is no longer needed.

For Ravid, surveillance is a tool that does not require reinventing management. “I don’t think the process of what it means to be a good manager—what it means to manage performance in a way that is transparent and just and fair—I don’t think that’s changed,” he said. Good management still requires consistent feedback, meaningful dialogue, and human judgment to interpret what the numbers show. “You need to have a discussion with an employee to talk about and contextualize what’s going on: Here’s what I’m seeing. Can you tell me about your perspective?”

Ethical monitoring is not possible, Ball argued, without businesses and managers prioritizing the human judgment that surveillance tools lack. “Managing with monitoring really requires that socially sensitive, socially supportive monitoring practice,” she said. The data are only as useful as the human relationship in which they are delivered. A manager focused purely on metrics, without the skills to contextualize them with their staff, is trading surveillance for the meaningful judgment and coaching they should already be providing. “It is up to organizations to be discerning, socially responsible, and strategic,” Ball said. “There are no quick fixes. Using these technologies well takes skill, but when done right, they can support both a healthy workplace culture and strong performance.”