OAKLAND, July 19 — A lawsuit filed in federal court here by 26 current and former Meta employees alleges the company used an artificial intelligence system to rank workers for the layoffs announced in May, without accounting for periods when employees were on parental or sick leave. The complaint, unsealed this week, paints a picture of a process that employees say was anything but the human-driven downsizing Meta executives described publicly.
Inside the courthouse documents, a stark discrepancy emerges. While company leaders told the public that decisions were made by managers, the complaint claims automated systems were secretly harvesting data on productivity, code modifications, performance evaluations, and usage of Meta’s own AI tools. The algorithm then scored each employee, determining which names would appear on the termination list.
‘Lacked context’
The AI system, according to the suit, detected drops in productivity among workers. Yet it lacked the context to understand why those drops occurred. An employee on parental leave, a software engineer sidelined by illness — the algorithm had no mechanism to recognize these absences. The complaint argues this led to unfair rankings and ultimately to people being laid off who would have remained employed had the company’s system accounted for legitimate leave.
One employee who took extended sick leave told investigators the algorithm flagged negative productivity metrics without any note of the medical absence. The lawsuit does not name individuals but describes a pattern of scores that penalized workers for time away from the keyboard — time that federal and state leave laws are meant to protect.
Damages and structural changes sought
The 26 plaintiffs — a mix of current and former employees — are seeking unspecified damages and a court order compelling Meta to change its layoff practices. Specifically, they want the company to stop using automated systems that fail to incorporate context about leave. They also want a review of the May 2026 layoffs to see if the AI’s scoring method violated employment law.
By late afternoon, representatives for Meta had not yet filed a response in public court records. But the case, now before a federal judge in Oakland, could force a broader reckoning over how tech companies deploy AI in human-resource decisions. The suit is one of the first to challenge the secret use of machine-learning tools to rank workers for termination — a practice that, if proven, may reshape how Silicon Valley handles downsizing.
For now, the algorithm’s role remains under legal scrutiny. The plaintiffs are asking the court to order Meta to turn over the source code used to score employees, along with internal communications about the layoff decisions. That discovery process, if it moves forward, could pull back the curtain on how — and how fairly — AI-driven management is really operating on the ground inside one of America’s largest tech employers.
