Somewhere right now, an employee is getting an automated flag for spending too long away from their keyboard. Another is being scored on keystroke volume. And another is having their Slack tone analyzed by an algorithm to determine whether they are a retention risk. None of this is science fiction. It is a normal day at a growing number of companies, and it is making everything worse.
I have spent more than 15 years researching workplace culture and employee well-being, and what I am watching unfold with AI-powered surveillance is one of the most self-defeating trends in modern management. The impulse behind it is understandable. Leaders want accountability, especially across distributed teams. But the execution is producing the opposite of what companies are hoping for, and the data is starting to prove it.
According to a study by ExpressVPN, 61% of companies now use artificial intelligence-powered analytics to track and evaluate employee performance, and WebsitePlanet finds that 84% of all documented AI surveillance cases have occurred since 2019. The pandemic normalized remote monitoring, and AI supercharged what employers could see. Today’s bossware doesn’t just log when you clock in and out. It tracks keystrokes, mouse movements, active application time, browsing history, email sentiment, facial expressions during video calls, and even the frequency and tone of your Slack messages. The monitoring infrastructure that companies have built in the past five years would have seemed extraordinary a decade ago. Now it is standard practice.
The examples keep coming, and some of them are genuinely remarkable. Meta workers were reportedly outraged after discovering internal software was tracking their keystrokes and mouse movements, with employees asking internally how they could opt out. Burger King deployed an AI chatbot to monitor whether employees say please and thank you during customer interactions. And in what may be the most vivid symbol of how far this has gone, Prego launched a dinner-table recording device so its pasta sauce brand could listen in on family mealtime conversations. That last one is a marketing stunt, not a workplace tool, but it captures something real about the cultural moment we are in: Recording and analyzing human behavior has become so normalized that it barely registers as strange anymore.
Here is the core problem with AI surveillance as a management strategy: It does not work. Not only does it fail to improve performance, but it also undermines the conditions that make people want to perform in the first place.
The WebsitePlanet research found that employees under surveillance were 1.5 times as likely to report poor mental health as those who were not monitored. Fully 56% of surveilled workers reported elevated stress levels. Perhaps most damning for the business case behind bossware: Harvard Business Review reported on a study that found that employees were more likely to break rules when they knew they were being monitored, not less, because surveillance diminishes their sense of personal responsibility and autonomy. When people feel watched rather than trusted, they stop behaving like owners of their work and start behaving like subjects of it.
The productivity theater this produces is real and widespread. A Software Finder study found that 66% of employees have stayed online or remained active after finishing their actual work to avoid appearing unproductive. Some of their most common tactics employees use include moving their mouse periodically (56%), keeping a decoy document or browser tab open (56%) and responding slowly to nonurgent messages to simulate activity (43%). Companies that deploy AI surveillance to catch slackers are instead creating an entire workforce of people optimizing for the appearance of work rather than the substance of it. They are measuring the wrong things and getting very good data on nothing that matters.
Stress and disengagement are bad enough. But the business cost that should concern leaders most is what surveillance does to retention. Surveillance damages the trust between employees and their employers, and trust is one of the strongest predictors of whether someone stays or leaves.
AI is giving employers dramatically more capability to monitor workers, including the ability to use behavioral data to train AI agents that can eventually perform the monitored tasks themselves. That framing, watching workers to eventually replace them with AI, is not lost on employees. When people suspect that the data being collected about them is part of a longer-term plan to automate their jobs, the trust damage is compounding. They are not just feeling watched. They are feeling set up.
I have watched this pattern play out across the research I have done with some of the largest employers in the world. The companies with the lowest turnover are not the ones with the tightest monitoring. They are the ones with the highest trust. Those two things are almost perfectly inversely correlated, and AI surveillance is pushing organizations in exactly the wrong direction.
None of this is an argument against accountability. It is an argument against confusing surveillance with accountability. They are not the same thing, and treating them as equivalent is costing companies more than they realize.
The organizations getting this right are measuring outputs, not inputs. They set clear expectations for what good work looks like, give people the tools and autonomy to do it, and evaluate performance based on results rather than keystroke counts. They use technology to remove friction and administrative burden from employees’ days, not to add a layer of ambient judgment on top of everything those employees do.
The deeper issue is one of philosophy. Surveillance assumes that people will not work unless they are watched. Trust assumes that people want to do good work and perform better when given the conditions to do it. Every company gets to choose which assumption it builds its culture around. The ones choosing surveillance are discovering, expensively and often too late, that the assumption was wrong.
AI has genuine and powerful applications in the workplace. Monitoring every bathroom break is not one of them. The companies that will attract and retain the best people over the next decade are not the ones with the most sophisticated bossware. They are the ones that figured out that trust scales better than tracking, and that the most productive thing a leader can do is get out of the way.