For five years, the former Meta employee had been feted as a high performer by her managers. She had even received a rare equity award in recognition of her performance, which had never fallen below an “exceeds expectations” rating.
Imagine her surprise, then, when she was unceremoniously laid off this past spring.
Perhaps she should have seen the writing on the wall. After she took a medical leave in late 2025, her performance rating slipped. It turned out she’d initially been given an “exceeds expectations” rating, but her manager had been forced to change her rating to fit a bell curve. Her job responsibilities also changed substantially. She had always ranked high on the internal dashboards Meta used to track AI adoption, which categorized its most avid practitioners as “AI first” or “AI native.” But after returning, she noticed her scores had dropped and was startled to realize the dashboards had not taken into account that she was on protected leave.
Another employee impacted by the layoffs, a former director at Meta, had come to a similar conclusion. He monitored his team’s usage of artificial intelligence closely, almost on a daily basis, and noticed the dashboards had “failed to distinguish employees’ leave from lack of engagement,” as he said in recent legal filings. In other words, there was no way to detect if someone’s ranking had been compromised simply because they were on leave.
In an era of forever layoffs, some tech workers have suspected that an algorithm or AI could be responsible for picking off employees. Now, it seems, those fears may be realized, as captured by a novel lawsuit that the Meta employees brought in July, alongside a number of former colleagues who were also laid off this year. The 26 anonymous plaintiffs made a bold claim: They accused Meta of using AI-enabled systems to determine which employees would be laid off in a major round of cuts this year—and that, as a result, the company had disproportionately targeted people who needed accommodations for disabilities or took parental leave.
Job displacement isn’t the only consequence of widespread AI adoption. A Fast Company LinkedIn poll reveals . . .
that the fears of being ousted by AI are palpable, and not just for disgruntled tech workers. More than two-thirds of people surveyed said they were worried about their company deploying AI to select employees for layoffs. Of more than 480 respondents, 37% said they were “very” concerned about the prospect of AI being used to that end.
For months, Meta has strenuously denied the claims laid out in the lawsuit, insisting humans were involved throughout the layoff selection process. In one declaration, Linh Doan, an HR leader, stressed that “AI did not select anyone for termination. AI did not assist anyone in making selection decisions.”
But those kinds of proclamations may be too little, too late. Workers know their employers already use automated tools to collect all kinds of data that informs hiring and promotion decisions. Major companies like JPMorgan Chase, Disney, and, of course, Meta are now enlisting AI to draft self-assessments and performance reviews. People analytics tools often track absenteeism and sick days, an exercise in productivity theater that employers nonetheless use to evaluate worker output—even though relying on that kind of data could harm people who take leave.
So it’s not hard to imagine that all this data could influence layoff decisions, as companies are armed with more information on their employees than ever before. Some managers and HR leaders have said as much in anonymous surveys, where they admit to using AI to trim head count. A handful of states have already passed legislation that will force companies to be more transparent about how AI factors into termination decisions, in anticipation of the potential impact on workers. Meanwhile, high-profile lawsuits against HR tech companies like Workday and Eightfold indicate that we are just beginning to understand how AI use could perpetuate discrimination in other stages of the employment cycle.
As AI becomes more pervasive across corporate America, it is increasingly difficult for workers to assess the extent to which it may be used in layoff decisions and how all the data companies are hoovering up could be wielded against them. Perhaps the real concern is not whether AI is conducting layoffs—but that there is so little visibility into how AI is radically transforming the way companies make decisions about their workforce.
The precedent for automating layoffs
As companies ramp up pressure on employees to adopt AI, and layoffs become a regular feature across workplaces, it seems only natural that the technology may start to influence how companies approach some of the most difficult decisions they have to make about their workforce.
“Making layoff decisions is an impossible problem because you are not only getting rid of people, but you’re also changing the structure of the work itself in ways that are very hard to predict,” says Richard Landers, a professor of industrial and organizational psychology at the University of Minnesota. “It makes using external tools very attractive [because] some of that responsibility is offloaded.”
While there is limited evidence that large employers are explicitly using AI tools to generate lists of employees to lay off, the idea of productivity metrics or automated systems driving employment decisions is hardly new.
“There has always been to a degree some form of measurement, explicit or implicit, that goes into layoffs,” says Joseph Fuller, a Harvard Business School professor of management. He points to the layoffs Ford conducted in the 1970s by asking supervisors which employees were most productive or spent more hours in the office—the type of metrics companies can now easily monitor through swipe card data or email traffic analysis. “Using AI to make those decisions is really in its infancy,” he says.
Warehouse workers are already subject to more monitoring and can automatically lose their jobs for not meeting productivity quotas at companies like Amazon. Gig workers at Uber and DoorDash can be automatically suspended or find their accounts deactivated with little human oversight; in fact, Uber is facing a fine of nearly $1 billion in the Netherlands for suspending drivers (though the company plans to appeal).
White-collar workplaces track such data points as sick days and other absences to measure absenteeism, even keeping tabs on disengaged employees who might be a “flight risk.” The COVID-19 pandemic and rise of remote work prompted some companies—among them JPMorgan Chase—to more closely monitor how their employees spent their time, tracking keyboard activity and meeting cadence to evaluate productivity. A New York Times report from 2022 found that eight of the 10 largest private employers in the country collected productivity data on their workers.
In 2025, the Department of Government Efficiency sought to automate layoffs across the government while making sweeping cuts to the federal workforce. A Reuters report revealed that DOGE revived an old piece of software called AutoRIF, which the government had developed decades before in an attempt to automate layoffs. The tool was deemed too “clunky” to actually use, according to Reuters.
But DOGE reportedly tapped a Tesla engineer to redesign the software to identify federal workers for layoffs, renaming it the “Workforce Reshaping Tool” and disseminating it to federal agencies for testing. (It’s not clear whether the software was involved in making cuts to the federal workforce—or whether AI was used for layoffs in any capacity.)
Multiple experts told Fast Company there are no tools on the market that are expressly designed to conduct layoffs, and in states like California, there are laws in place to govern how companies use AI or automated systems for employment decisions. But that could change soon.
A VC funding flood
In the first quarter of 2026, AI companies reportedly received more than 80% of overall venture capital funding—something like $242 billion—which means plenty of new tools could emerge on the scene. And some companies may simply position their tools in a manner that evades scrutiny. One researcher told Fast Company they had recently approached a people analytics provider to partner on a study examining AI-driven layoffs. The company—which the researcher described as a market leader—ultimately declined to participate because its software and AI-driven products were being branded as workforce management tools, with the implicit acknowledgement that “workforce management is code for layoffs.”
Once AI becomes embedded in virtually every workflow, it is perhaps unavoidable that termination decisions will draw on inputs from automated systems. In the near future, asking whether AI was involved in the process, as Fuller puts it, “will be the equivalent of saying: ‘Did you use any data in this decision?’”
As workforce management tools grow more sophisticated, companies will likely turn to AI and automation to evaluate individual employees in the context of layoffs. One way employers are already using AI to make cuts, however, is in the context of broader changes to their workforce as they eliminate job functions “in one fell swoop,” according to Fuller.
“Companies are beginning to understand what the first generation of agents can do, and that’s causing them to start adjusting workforces,” he adds. “It’s not so much that they’re doing the layoff because they’ve analyzed individual workers’ performance using AI. It’s because they’ve made a judgment about the durability of an existing job description in light of AI.”
The influence of ‘work intelligence tools’
Some employers are more and more interested in what the Gartner analyst Emily Rose McRae describes as “work intelligence tools,” which are effectively designed to meet this current moment. McRae says these tools claim they can identify which jobs will be most exposed to AI at a given company. “The phrase usually used is ‘impacted,’ or where there is ‘opportunity,’” she explains, “by which they mean where they think you could replace head count.”
In practice, however, these tools—which McRae says some companies are using to determine which jobs should be cut—are not very good at taking into account all the variables that influence whether a job can truly be supplanted with AI, echoing the concerns many economists have expressed.
“It’s appealing to have an external party say, ‘Here’s what the impact will be,’” she says. “It doesn’t feel as personal.”
Large employers tend to have access to more robust tools—whether developed internally or acquired through third-party vendors—as well as legal teams that should, in theory, ensure layoffs don’t inadvertently target specific groups of employees. The concern at smaller or midsize companies is that managers may get limited guidance on layoff decisions or lack access to enterprise grade tools, leaving them to seek out informal help from chatbots like Claude or Copilot.
‘A veneer of objectivity’
Hatim Rahman, a management and organizations professor at Northwestern University, is not surprised to hear that managers are using AI or related tools to make difficult termination decisions. “AI and algorithms give a veneer of objectivity,” he says. “Numbers can be very seductive in providing justification for why certain people can be laid off. The problem is that many times the people who are making decisions about layoffs don’t have an understanding of how the AI tools or workforce analytics that are being used are generating those recommendations or that data.”
If a hiring manager inputs performance review data and asks for layoff recommendations, for example, the chatbot may not consider whether the reviews are capturing the right metrics in the first place. “The thing that’s new is that anybody can do this, which means they don’t have to have any understanding of human resources principles or the relevant laws,” says Tara Behrend, a professor at Michigan State University’s School of Human Resources and Labor Relations.
“The question is: Do you have the right data?” she adds. “Are you making the decisions in a fair way based on things that actually matter, or are you picking up on noise and just sort of spurious nonsense? The recent advances in AI just make it even less apparent whether the data are meaningful—whether they’re picking up on signal or noise.”
‘Some red flags’
When researchers at Rutgers University in New Brunswick, New Jersey, started looking into how AI can shape layoffs at small and midsize companies, they hoped to better understand how reliably chatbots like ChatGPT and Gemini acted when selecting people for layoffs. But even with an engineered prompt, the results were not consistent, according to TJ Raineri, a PhD candidate and one of the researchers working on the study. “There were some red flags that we saw, including overindexing on laying off women and people of color,” he says.
In a subsequent study, Raineri and another researcher, Yixiao Shen, are looking into why this gap exists, and how those results compare to when a human oversees the selection process. “Big companies are getting all the attention, but midsize companies are doing layoffs,” Raineri says. “They might not be on the same scale as Amazon or Meta, but they’re doing layoffs, and I think we need to understand [that].”
Using chatbots or other AI tools could be especially appealing to managers who have never overseen job cuts and might not trust their own instincts. But an overreliance on AI can end up reshaping how they approach those decisions—a concept that Landers, the University of Minnesota professor, describes as abdication to AI. “Over time, a person who is accustomed to making decisions with AI input gradually is likely to find that their decision-making processes have not just been influenced but potentially replaced with that of AI without their knowledge,” says Landers. “And the biases that [those systems] exhibit will essentially creep their way into their judgment without them realizing it.”
The workers who pay the price
The recent lawsuit has forced Meta to disclose an unusual amount of information about how the company conducts layoffs and what selection criteria are taken into consideration. Among those criteria were performance ratings and job profiles, both of which are now being influenced by AI’s growing incursion into workforce management.
The suit also invites uncomfortable questions about how the growing adoption of AI could unintentionally target certain employees, especially as companies become leaner. Perhaps most concerning is that there is so little clarity on how, exactly, bias can be magnified by automated systems—as well as limited insight into how employers arrive at layoff decisions—leaving workers with scant opportunities to raise concerns.
Meta has disputed the claims put forth by the plaintiffs, and the company has explicitly denied that managers took leave history or disability status into consideration during the layoff selection process. But as the employment attorney Christine Webber points out, this could actually make managers more likely to sign off on a termination decision that might have required closer scrutiny. “Their claim of the managers not knowing actually makes it all the more likely that they would rubber-stamp something that really was driven by those excused absences,” says Webber, a partner at Cohen Milstein and cochair of the firm’s civil rights and employment practice.
Even when companies ensure humans are in the loop on critical decisions, they might be implicitly cosigning recommendations made by AI, especially when they’re being tasked with reviewing vast amounts of data. “What I’ve seen commonly in cases involving reliance on AI is this claim that it’s not really making the decisions,” Webber says. “But then, if you look at how quickly we go from the AI’s recommendation to the final decision, there’s not really a lot of room in there for the managers to have done much.”
‘An invisible cage’
As the use of AI and algorithms in the workplace has been normalized, employees are being confined by an “invisible cage,” according to Rahman, the Northwestern professor. This cage effectively controls their “opportunities, success, and layoffs,” but they still have little transparency into how those systems actually work. “It’s that information asymmetry for what data is being collected, how it is being processed, and how it is being used,” he says. “And I think it’s throughout—it’s from hiring and performance to layoffs.”
A few states already have laws in place to address exactly this issue. In addition to a new law in California that creates guardrails around how companies use AI to execute layoffs, both Illinois and Colorado now require companies to disclose when AI dictates a termination decision. In New York, employers who submit WARN notices—federally mandated advance notifications of mass layoffs—must also disclose if a layoff is due to AI adoption.
Employers may have little incentive to disclose how they navigate layoffs beyond what is legally required, but they may also not be particularly well served by offering so little transparency. In the absence of information, workers will fill in the gaps—and likely assume the worst.
“Workers will come up with their own theories of how decisions are being made,” Rahman says. “Organizations that can find the right balance in terms of providing transparency without encouraging gaming or those types of behaviors will hopefully generate more trust and long-term commitment from workers.”
Retaining humanity
Without more transparency, one of the few ways workers can gain insight into how companies operate—and possibly appeal decisions they suspect did not adequately involve human judgment—is through lawsuits like the one brought against Meta. As business leaders urge them to embrace AI and tout sizable investments in the technology, employees have ample reason to question how companies are making consequential decisions about their workforce.
“I think we have to have humanity at the center of all of these decisions,” says Landers. “The lack of human values around the edges of these kinds of decisions is really degrading the experience of work for everyone, which I think we’re feeling right now. . . . One of the reasons why there’s such a mass negative reaction to AI is because most of the use cases are actively removing the humanity from the equation.”