Skip to main content

labor ‘Constantly Monitored’: The Pushback Against AI Surveillance at Work

Academics decry algorithmic monitoring of workers and call for stronger standards – but U.S. unions have been slow to act.

Hugo Herrera, Stanford Social Innovation Review

From algorithms firing staff without human intervention to software keeping tabs on bathroom breaks, technologies including artificial intelligence are already upsetting workers and unsettling workplaces.

At call centers, AI systems record and grade how workers handle calls, often giving failing grades for not sticking to the script. Some corporate software spies on workers to see whether they ever write the word “union” in their emails.

As technologies grow ever more sophisticated in monitoring, surveilling and speeding up workers, many workplace experts say US businesses, labor unions and government are not doing nearly enough to protect workers from tech’s downsides.

“Workers are being constantly monitored, and AI-based monitoring tools can make mistakes that can translate into unfair pay cuts or firings,” said Virginia Doellgast, a professor of employment relations at Cornell. “Workers often don’t know what monitoring tools are being used, what data the tools are collecting or how that data is used to evaluate their performance.”

In Europe, unlike in the US and Canada, many unions have been pushing for years for protections against some of the more intrusive ways that AI tools track and manage workers. “This issue has yet to be put at the center of the radar for unions in North America,” said Valerio De Stefano, a labor law professor at York University in Toronto, who has written extensively on AI’suse in the workplace. “Unions in Europe are more aware of the uses of technologies from the surveillance standpoint. This is not something that unions in North America have focused on.”

At some German companies, labor experts say, workers have won protections that could become models for US and Canadian workers. At Deutsche Telekom, Germany’s largest telecommunications company, workers have won a prohibition against algorithms firing workers without any human involvement as well as a ban on using data collected by digital monitoring to discipline or dismiss individual workers.

“In Europe, workers have stronger rights to obtain information and participate in decision-making,” Doellgast said. “In the US, where there is a union present, workers have some information rights about AI and hopefully a voice in how it’s used. Where unions aren’t present, workers have no information rights, and all they see is the effects of the technologies on them.”

Mindful of the downsides of AI and algorithmic management, US labor unions are starting to push harder for protections. For instance, at some call centers, the Communications Workers of America union has won not only requirements that managers notify workers whenever recording their calls, but also guarantees that management will only record calls for training purposes to help improve employee performance – and not for evaluating or disciplining workers.

Dan Reynolds, the Communications Workers’ assistant research director, said the union has long been concerned about how new technologies affect jobs. “AI is a new technology often used to speed up the work, deskill the work, make workplaces more stressful and make jobs more demanding,” he said. “Our concern about AI isn’t just its effect on the number of jobs, but how it will affect the quality of jobs.”

If you like this article, please sign up for Snapshot, Portside's daily summary.

(One summary e-mail a day, you can change anytime, and Portside is always free.)

“Our goal is not to stop new technologies,” Reynolds continued, “but to make sure the gains of these new technologies are broadly and equitably shared.”

Germany has laws requiring companies to notify their works councils about AI and other new technologies that they’re planning to adopt. Most German companies have work councils, which are worker-management committees that discuss everything from vacation schedules to the pace of work and the effects of AI.

To prevent AI-based performance data from being used against individual workers, the work councils at Deutsche Telekom got the company to agree that performance data can be gathered only for groups of at least five employees. The company has also agreed not to use AI to gather certain personal information about employees, such as their political opinions or sexual orientations.

Doellgast and De Stefano – who edited a recent academic journal about AI and work – say worker input about new technologies often reduces their invasiveness and other downsides for workers, while making the introduction of technologies smoother and more productive. Moreover, when workers have some say about new technologies, that often reduces employee resistance to those technologies.

De Stefano pointed to some problems with using AI to hire and discipline workers. “These machines are, in many cases, unreliable,” he said. “They have certain discriminatory output, especially in hiring. These machines are basically benchmarked around a standard worker – normally white, prime-age, male workers. Anyone who doesn’t correspond to that benchmark risks being misjudged by these algorithms.”

The AFL-CIO, the main US labor federation, has created a technology institute to develop expertise and policies on AI and other technologies. That institute is planning training sessions to educate union leaders and strategists about new technologies.

“In sectors where performance monitoring and algorithmic management are present, you can have a lot of negative impact,” said Amanda Ballantyne, director of the AFL-CIO’s technology institute. “If you dive into a sector like Amazon, many workers wear wearables that track every movement their body makes, everywhere they go, how fast they complete tasks, how long they’re off task, how long they take in the bathroom. It’s [scientific management] on steroids.”

Ballantyne said the communications workers union, the hotel workers union, Sag-Aftra and the Writers Guild have helped lead the way on new technologies. After their recent 148-day strike, the Writers Guild won protections that require studios to disclose whether any material given to writers was developed with the help of AI.

The US Chamber of Commerce says AI and analytics can have substantial benefits for workers and productivity, providing insights into worker performance and allowing for targeted coaching and training to improve performance.

AI surveillance, the chamber says, can also help prevent workplace violence by, for instance, monitoring abnormal behavior in the workplace. While many workers criticize the use of “wearable” monitors, the chamber says “smart sensors and wearable devices can help” protect them, by detecting potential accidents, ergonomic risks, toxic chemicals and imminent heat stress.

“While there are clear benefits” to AI, said Michael Richards, policy director of the chamber’s technology engagement center, “we understand there are legitimate concerns surrounding the use of the technology.”

Employers “recognize that engaging in an inclusive dialogue about using new technologies is critical to fostering a culture of trust with employees”, he added.

Annette Bernhardt, director of the technology and work program at the UC Berkeley Labor Center, pointed to the home-care sector as an example where new technologies make workers’ lives more stressful. Many home-care aides must meticulously follow instructions that apps send them, while reporting back each and every task they complete.

“We need strong labor standards around the use of these technologies,” Bernhardt said. “We need to support unions as they bargain around these technologies. Most important, we need to assure that workers have a seat at the table on these technologies from the outset, not just when they’re being implemented.”

Bernhardt added: “When workers are at the table, it means better technology adoption for them and better results for employers.”