Artificial intelligence is entering the heart of corporate decisions, but this time not to create new products or automate processes. According to 26 former and current Meta employees, it was used to decide who to lay off. The lawsuit filed in the United States accuses the role of algorithms in the company’s mass layoffs, alleging that the systems adopted ended up penalizing workers on medical, parental leave, or with disabilities.
THE LIST OF LAYOFFS UNDER SCRUTINY
The lawsuit was filed in the federal court in Oakland, California, by 26 employees who requested anonymity. The workers are part of about 8,000 employees, equal to 10% of Meta’s global workforce, involved in the staff reduction plan announced by the company in May. According to the communication received, the layoffs were supposed to take effect from July 22.
At the center of the complaint is the method Meta allegedly used to select employees to be let go. The former and current employees involved in the lawsuit claim that the company did not leave the decision solely to managers but used a series of internal tools based on artificial intelligence to evaluate performance, productivity, and work activity.
“Meta did not compile the layoff list through the thoughtful judgment of managers who knew the work,” the complaint reads, according to Courthouse News. On the contrary, the company allegedly used systems capable of “assigning scores, ranking, and selecting employees to include on the list.”
THE AI TOOLS IN THE CROSSHAIRS
The complaint claims that Meta used several internal systems to classify workers. Among these are Metamate, an assistant based on a large language model, systems called “second brain” trained on employees’ documents and communications, and productivity assessment tools powered by data such as keystrokes, screen activity, emails, browser history, and AI token consumption.
It is precisely this last parameter, reports Reuters, that the plaintiffs point to as one of the most controversial elements. Token consumption has become an indicator of AI tool usage, but according to the lawyers, it can become a penalizing criterion for those who, for legitimate reasons, could not perform the same amount of activity recorded by the system.
The problem, the workers argue, is that an algorithm can record an absence as a reduction in productivity without necessarily understanding the reason for that absence.
THE RISK OF PENALIZING THOSE ON LEAVE
The core of the accusation concerns employees who had taken protected leaves. The lawsuit states that Meta’s systems used data that, by their nature, were lower for those who had been away from work for medical, family, or pregnancy-related reasons.
“These tools rely on data such as performance evaluations, calibration scores, productivity and output indicators, ‘AI-native’ assessments, and AI token consumption, elements that cannot be accumulated by an employee who is on protected medical or family leave,” the lawyers wrote in the complaint.
Among the cases cited, writes the Guardian, is that of a researcher who received the layoff notice two days before giving birth, an engineer who claims to have received a lower evaluation after a period of absence due to injury, and a manager laid off after 16 days of medical leave.
DISCRIMINATION ALLEGATIONS
The plaintiffs claim that Meta violated several U.S. laws, including the Americans with Disabilities Act, the Family and Medical Leave Act, the Pregnancy Discrimination Act, and the Pregnant Workers Fairness Act.
The lawsuit also invokes the principle of so-called “disparate impact,” according to which a seemingly neutral practice can be discriminatory if it disproportionately affects a protected category. According to the employees’ lawyers, a system that negatively evaluates periods of absence can impact women more, who more frequently take leaves for pregnancy and family care.
The workers also accuse Meta of not subjecting its AI tools to sufficient checks to identify possible biases, contrary to new rules introduced in California and New York City on automated decision-making systems.
META’S RESPONSE
The company denies the allegations and claims that layoff decisions were made by people, not algorithms.
The matter arises as Meta is profoundly transforming its organization around artificial intelligence. The May cuts were presented as part of a broader reorganization, accompanied by the shift of thousands of employees to new AI-dedicated teams.
However, the lawsuit raises a broader issue: when algorithms enter workplace decision-making processes, who controls the criteria by which people are evaluated and who is responsible for any discriminatory effects? For the employees who took Meta to court, the answer now lies with the judges.
