26 former Meta employees sue, claiming AI tools scored and ranked them for termination while they were on medical and parental leave

When the Algorithm Picks Who Gets Fired

Meta says humans decided. The lawsuit says an AI did the deciding and humans just signed off. That distinction is about to reshape how every company uses AI to manage its workforce.

Twenty-six former Meta employees filed suit in the US District Court for the Northern District of California, alleging the company deployed a "constellation" of internal AI systems to score, rank, and select employees for termination during May layoffs that cut roughly 10% of staff, or about 8,000 workers 1.

Reuters called it "the first against a major US company to challenge the alleged use of AI in conducting layoffs" 2. That framing matters. This is not a labor dispute with an AI subplot. It is the opening case in a new category of litigation.

The tools named in the complaint

The lawsuit describes a stack of AI systems that allegedly fed into termination decisions. The list is specific: Metamate, Meta's internal AI assistant; custom agents that employees had trained as personal productivity systems; dashboards tracking how much AI each worker consumed; activity and keystroke monitoring; and automated performance scoring layered on top of all of it 2.

Workers were slotted into adoption tiers based on their AI usage, with labels like "AI Native," "AI First," and "AI Enabled" 2. The problem is structural: those metrics reward activity and tool engagement. Someone on medical or parental leave generates none of it. While they are away, their scores flatline, and the ranking system reads that flatline as poor performance.

The complaint accuses Meta of failing to "neutralize those inputs" for employees on protected leave 2. In practice, that meant workers who took leave kept getting scored on metrics they had no chance to produce while away. Nobody removed them from the candidate pool. Nobody hit pause for the kind of individualized review that employment law expects before you fire someone. The system ranked them lower, flagged them as underperformers, and slotted them onto the termination list. The plaintiffs say the effect was punishment for doing something the law explicitly permits 2.

The complaint describes one plaintiff, a Meta scientist, who received her termination notice while on approved pregnancy leave. The timing: the day before her water broke, two days before she gave birth 2. All 26 plaintiffs had requested leaves or disability accommodations in the 24 months before being selected 2.

Meta's defense, and why it matters more than the denial

Meta spokesperson Tracy Clayton offered a flat denial. "These claims lack merit and are not based on facts," Clayton said. "Workforce management and organizational decisions were and are made by people, not AI" 1.

That denial is the whole case. If Meta can establish that a human reviewed and approved each termination, the company has a strong defense under existing labor law. The question is what "made by people" actually means when the ranking, scoring, and shortlisting were done by algorithms. A manager who signs off on a list generated by an AI scoring system has technically made a decision. Whether that satisfies the law is unresolved.

This is the legal gap the case will probe. If Meta prevails, the "humans signed off" defense becomes a template. Any company could deploy AI to rank, score, and shortlist employees for termination, then claim humans made the final call. If Meta loses, every AI tool touching hiring, performance review, or termination decisions becomes a litigation liability overnight.

The regulatory backdrop

The complaint invokes four federal statutes: the Family and Medical Leave Act, the Pregnancy Discrimination Act, the Americans with Disabilities Act, and the Pregnant Workers Fairness Act 2. California's Fair Employment and Housing Act also factors in, with its prohibition on automated-decision systems that produce discriminatory outcomes tied to disability or pregnancy 2.

The case is not a class action. Meta's employment agreements require individual arbitration, which waives the right to participate in class actions 2. The plaintiffs are seeking a court order to freeze their terminations, scheduled for July 22, pending an independent review of how the algorithm made its picks 2. That audit would examine what data the system relied on and how heavily each factor was weighed. It would also "recompute selection scores using leave- and accommodation-neutralized inputs" to determine whether any plaintiff would have kept their job had leave status been excluded from the calculation 2.

All of this happened against an odd financial backdrop. Meta had posted record revenue the month before and was on track to spend somewhere between $125 billion and $145 billion on AI for the year 2. The same technology absorbing that investment may have selected which employees would not survive to see its returns.

So what

For builders: any HR AI tool that feeds into employment decisions now carries legal exposure. If your system produces rankings or scores that flow into termination decisions, you need to prove it accounts for protected leave status. Otherwise you are building a liability engine.

For investors: enterprise AI valuations assume smooth deployment. This case introduces a risk variable few models price in. If courts rule that algorithmic employment decisions create employer liability, the compliance overhead for HR AI tools could constrain the addressable market for the entire category.

For users: your job security may already be scored by a system nobody fully understands. The tools named in this lawsuit exist in some form at most large companies. The question is not whether your employer uses them. It is whether anyone checked if the scoring works.