Did an AI Decide Who Got Fired? Meta Faces Landmark Lawsuit Over Layoff Algorithms

Meta is heading to court over a question that could reshape how every company uses artificial intelligence in the workplace: Did an algorithm decide who lost their job? A new federal lawsuit filed Monday by 26 current and former Meta employees alleges that the company deployed a “constellation of internal artificial-intelligence systems” to score, rank, and select roughly 8,000 workers—10% of its workforce—for a mass layoff announced on May 20. The plaintiffs claim those AI systems disproportionately targeted employees who had taken protected medical, family, disability, or pregnancy-related leave, effectively penalizing them for exercising their legal rights. Meta denies the allegations, insisting that “workforce management and organizational decisions were and are made by people, not AI.”

The lawsuit, filed in the U.S. District Court for the Northern District of California in Oakland, focuses on Meta’s most recent large reduction in force. The company began notifying employees on May 20 about the layoffs, which are scheduled to take effect on July 22. Notably, many of the 26 plaintiffs are still on Meta’s payroll, with their separations set to begin later this month—a detail that makes the case unusually urgent and has prompted the plaintiffs to seek a preliminary injunction to block the layoffs while they pursue discrimination claims in private arbitration.

The complaint alleges that Meta used a suite of internal AI tools to generate productivity and performance scores that directly influenced who ended up on the layoff list. Among the systems named are Metamate, an internal large language model assistant; “second-brain” agents that ingest workers’ communications and documents to replicate or track their output; keystroke and activity-monitoring tools that track screen content, emails, and browser history; AI-token-usage dashboards that measure how employees use AI tools; and algorithmically assisted performance rankings and calibration systems. The plaintiffs argue that these systems draw on inputs—performance ratings, calibration scores, productivity metrics, “AI-native” ratings, and AI-token consumption—that “by design, cannot be accumulated by an employee who is on protected medical or family leave, or whose output is reduced by a disability.”

This is a first-of-its-kind employment case focusing on AI-assisted layoff targeting. While courts have long dealt with algorithmic bias in hiring (e.g., resume screening tools), this lawsuit pushes the frontier into termination decisions—and specifically into the use of AI to measure productivity in ways that may inadvertently discriminate against workers on protected leave. The case directly intersects with emerging regulatory frameworks. The plaintiffs cite recently adopted AI-related laws in California and New York City that require companies to test their AI systems for bias, and they allege Meta failed to do so.

The lawsuit lands at a moment of intense legal and public scrutiny for Meta. The company recently suffered major court defeats in other matters, including a $375 million verdict in Santa Fe, New Mexico, and a $4.2 million judgment in Los Angeles over platform safety and mental health impacts. While those cases are unrelated to AI layoffs, they paint a picture of a company already under a microscope. This new case could force Meta to open the black box of its internal AI tools in discovery, revealing how the company actually manages and evaluates its workforce.

For any company using AI in workforce management, this case is a wake-up call. The lawsuit invokes multiple federal and state laws, including Title VII’s disparate-impact doctrine, the Family and Medical Leave Act, the Americans with Disabilities Act, the Pregnancy Discrimination Act, and the Pregnant Workers Fairness Act. The allegation is not that Meta intended to discriminate, but that its AI systems had a disproportionate adverse impact on protected groups—a classic disparate-impact claim, now applied to algorithmic decision-making.

The practical implication is clear: companies that deploy AI tools to score, rank, or monitor employees must audit those systems for bias, especially when the metrics used (like keystroke counts, AI-token usage, or output volume) are inherently biased against workers who take leave or have disabilities. The plaintiffs argue that Meta’s systems failed to account for the fact that an employee on protected leave cannot generate the same productivity data as a fully present colleague. If the court agrees, it could force companies to redesign their AI tools to exclude or adjust for leave-related data gaps.

The immediate focus is on the plaintiffs’ request for a preliminary injunction to halt the July 22 layoffs. A ruling could come any day. If granted, it would temporarily freeze Meta’s reduction in force and force the company to defend its AI systems in arbitration. Even if denied, the case will proceed to discovery, where Meta’s internal AI tools will likely be scrutinized in unprecedented detail. The outcome could set a precedent that echoes far beyond Menlo Park, influencing how every company—from startups to Fortune 500s—uses AI to manage, evaluate, and terminate its people.