Twenty-six Meta workers say the company’s own artificial intelligence tools turned protected family and medical leave into a layoff risk.
Story Snapshot
- Meta faces a federal lawsuit from 26 current and former employees over recent layoffs tied to AI scoring systems.
- The workers claim internal artificial intelligence and activity tracking punished people on medical, maternity, and parental leave.
- Meta says humans made the decisions and the claims “lack merit,” setting up a direct clash over how AI is really used.
- The case could shape how every large employer uses data and algorithms when deciding who stays and who goes.
Workers say Meta’s AI turned time off into a liability
A group of 26 Meta employees filed a lawsuit in federal court in Oakland, California, saying the company used artificial intelligence systems to pick who would be laid off from a major round of job cuts.
They are among about 8,000 workers, roughly 10% of Meta’s workforce, who were told their jobs would end in May as the company continued its push for “efficiency” and cost cuts. The plaintiffs say the way Meta used AI made protected leave an invisible strike against them.
A lawsuit accusing Meta of discriminatory artificial intelligence use in firing workers demonstrates how companies leaning on the emerging technology for employment decisions can incur liability. https://t.co/jpeh9SPay7
— Bloomberg Law (@BLaw) July 28, 2026
The complaint describes internal tools that sound like something out of a Silicon Valley pitch deck, not a human resources office. Reports say Meta relied on a mix of keystroke and screen-activity monitoring, code output, use of AI assistants, and algorithmically boosted performance rankings to build layoff lists.
One system, called “Metamate,” is described as an internal large language model assistant that acts like a “second brain,” tracking communications and documents across the company.
Protected leave meets productivity dashboards
The heart of the workers’ claim is simple: you cannot rack up points in a productivity system when you are not supposed to be working.
The lawsuit says scores and ratings built from keyboard data, emails, browser history, and other activity “by design, cannot be accumulated by an employee who is on protected medical or family leave, or whose output is reduced by a disability.” That design choice, they argue, turned lawful leave and disability accommodations into quiet penalties inside the layoff formula.
The complaint ties the alleged harm to specific protected categories. About half the plaintiffs reportedly took leave for caregiving or pregnancy-related reasons.
Eight are women who used maternity or pregnancy leave, four are men who took parental leave, and one woman took leave to care for a family member and later used bereavement leave.
Others had medical conditions or disabilities and say accommodation needs lowered their measured activity. When Meta looked at raw scores, the lawsuit claims, it did not adjust for any of this.
The human cost behind the numbers
Behind every metric is a person, and some of the stories in the lawsuit are jarring even for hardened readers. One plaintiff reportedly received a layoff notice while she was on approved pre-birth leave, just two days before giving birth, after planning her leave around company rules.
Another worker with a disability says Meta offered no accommodation, which meant his activity looked low long before any layoff decisions. These are not edge cases in the complaint; they are used to show how cold tracking systems can feel when real lives are at stake.
The workers are not only asking for money after the fact. They are seeking a preliminary ruling from the court to block Meta from completing the layoffs while they press their claims in private arbitration.
That request for a quick injunction says they believe the harm is not just past tense but real and present. So far, a judge has declined to stop the layoffs, which does not decide the core discrimination issues but shows how hard it is to win fast relief when proof sits mostly inside company servers.
Meta’s defense and the fight over proof
Meta has pushed back hard, and in plain language. A spokesperson said the claims “lack merit and are not based on facts” and insisted that “workforce management and organizational decisions were and are made by people, not AI.”
That statement hits two notes that matter for many readers: first, that this is a routine business restructuring, and second, that it is not some rogue robot firing people on its own.
The legal fight will turn less on slogans and more on records. The workers accuse Meta of breaking federal and state laws that ban discrimination or retaliation against workers who have disabilities, take medical leave, or are pregnant.
They also say Meta failed to test its systems for bias, which they argue violates new rules in places like California and New York City that demand checks on automated tools.
To prove this, they will likely need internal model outputs, ranking spreadsheets, and emails showing how managers treated low scores for people on leave.
Why this case matters far beyond Meta
This is not the first time software has helped drive layoffs, but it appears to be the first lawsuit against a major United States company that squarely challenges alleged use of artificial intelligence in mass job cuts. That “first” label is not just media hype.
If courts say companies must adjust AI-driven metrics for protected leave, every large employer who tracks keystrokes and dashboards will need to rethink their systems or risk similar claims.
The broader pattern in modern employment law is now clear. Companies turn to data, scoring, and algorithmic tools to make hard choices look neutral and efficient. Workers who feel singled out point to how those systems treat protected status, like pregnancy, disability, or family caregiving.
When employers control all the logs and code, it becomes hard for workers to show a direct link from score to pink slip. That proof problem, more than the buzzword “AI,” may decide how this case ends.
Sources:
abc7.com, reuters.com, youtube.com
