Twenty-six current and former Meta employees have sued the company, alleging that it used a “constellation of internal artificial-intelligence systems” to select workers for its spring layoffs in a way that discriminated against people with disabilities and those on medical, family or pregnancy leave.
The lawsuit, filed anonymously on July 13 in federal court in Oakland, California, says the plaintiffs were among the roughly 8,000 workers, about 10% of Meta’s global staff, cut in the May round, with their terminations set to take effect July 22.
They are asking the court to block the layoffs from being finalized and to order an independent audit of the algorithmically assisted selection process while they pursue their claims in private arbitration. Meta denies the allegations, telling CNBC the claims “lack merit” and that “workforce management and organizational decisions were and are made by people, not AI.”
The central allegation concerns which metrics the systems weighed. According to the complaint, Meta scored and ranked employees using performance ratings, calibration scores, productivity and output metrics, “AI-native” ratings, and internal records of AI-token consumption, a measure of how much an employee used AI tools that has become a proxy for productivity inside the company.
The plaintiffs argue these inputs “by design cannot be accumulated by an employee who is on protected medical or family leave, or whose output is reduced by a disability.” In other words, a worker physically absent on approved leave would automatically register as less productive and rank lower, regardless of their actual performance. The suit names internal tools including an assistant called Metamate and algorithmic scores allegedly drawn from data like keystrokes, browser history and email activity.
The plaintiffs, who come from six states and the District of Columbia, accuse Meta of violating the Americans with Disabilities Act, the Family and Medical Leave Act and the Pregnancy Discrimination Act, and of failing to test its AI systems for bias as required under recently adopted California and New York City laws. They also allege that Meta was made aware of the problem and did not correct it. Meta cut the jobs partly to offset the hundreds of billions of dollars it is committing to AI, even as CEO Mark Zuckerberg has said he expects no further company-wide layoffs this year.
A Test Case for Algorithmic Firing
The suit is one of the first major US cases to squarely ask what legal responsibility a company bears when algorithms shape mass layoffs. Its core theory is disparate impact: even a facially neutral metric like token consumption becomes discriminatory if it systematically disadvantages a protected group, here by penalizing the very absences the law protects.
The token-usage angle is especially pointed, turning a metric Meta reportedly used to signal AI adoption into alleged evidence of bias, and it cuts against the broader industry narrative that heavy AI use is simply a marker of a productive employee. If the plaintiffs win even a preliminary injunction, it could force companies to prove their layoff algorithms account for protected leave, reshaping how AI is used in workforce decisions.
The Wider Legal Reckoning
The case does not stand alone. It follows a June ruling in which a California federal judge held that the software firm Workday must face claims that its AI hiring tools discriminated against applicants, a decision signaling courts are willing to scrutinize algorithmic employment systems. Meta also faces a separate age-discrimination suit over its 2025 cuts.
Together these cases mark a shift from abstract worry about AI and jobs to concrete legal accountability, testing whether “the algorithm decided” can shield an employer from liability. Meta’s defense, that humans made the final calls, will likely be central, since the law generally holds that a company remains responsible for a discriminatory outcome no matter what tool produced it. The outcome could set an early precedent for the fast-growing use of AI in hiring and firing, at a moment when such tools are proliferating faster than the rules governing them.
