Workers Say Meta Used AI to Target Medical and Parental Leave in Mass Layoffs

A lawsuit puts workplace AI on trial

Twenty-six Meta employees have sued the company in federal court in California, alleging that internal artificial-intelligence systems helped select workers for a mass layoff and disproportionately harmed people taking protected medical, parental or family leave. The complaint concerns Meta’s announced reduction of roughly 8,000 jobs and seeks to halt terminations scheduled to begin July 22.

The allegations are not proven findings. Meta rejects them, saying workforce and organizational decisions were made by people rather than AI. That dispute is precisely why the case matters: courts may have to examine where algorithmic scoring ends and human responsibility begins.

How the alleged system worked

The employees say Meta relied on a collection of tools including algorithmically assisted performance ratings, activity and keystroke monitoring, productivity measures and dashboards tracking the use of AI tokens. According to the complaint, those signals were used to score and rank employees for inclusion on layoff lists.

The central criticism is mathematical as much as legal. Workers on approved leave cannot accumulate activity, output or tool-usage metrics during their absence. If a model treats missing activity as weak performance without correcting for protected leave, a formally neutral score can reproduce discrimination.

Human decision or automated influence?

Meta’s position is that managers made the decisions. But human involvement does not automatically settle the question. An algorithm can shape a shortlist, define performance categories or direct managers’ attention even when a person signs the final decision.

The employees are asking for an independent audit of the tools and for leave-related gaps to be removed from performance evaluations. They also seek reinstatement, back pay and other relief. A court has not yet determined whether Meta used AI as alleged or whether any system caused unlawful discrimination.

Privacy concerns widen the controversy

The complaint also describes workplace monitoring that captured signals such as keystrokes, mouse activity, browser history and communications on company devices. A related program was paused after more than 1,600 employees reportedly signed a petition raising privacy concerns.

Monitoring data can be useful for security or operational analysis, but repurposing it for employment decisions creates a different risk profile. Employees may not know which data matter, how errors are corrected or whether absence, disability and different working styles are fairly represented.

Why this case could travel beyond Meta

Companies increasingly use AI in recruitment, scheduling, productivity analysis and performance management. Employment decisions are consequential: they affect income, health coverage, immigration status and family stability. The case illustrates why “a human was involved” may be an inadequate governance standard if the human relies on opaque rankings.

A stronger model would require documented purposes, bias testing, leave-aware data handling, meaningful human review and a process allowing employees to challenge errors before harm occurs.

What remains unknown

The public record does not yet reveal the systems’ source data, model weights, error rates or the exact authority given to managers. The plaintiffs’ examples are serious but do not by themselves prove a company-wide pattern. Meta will have an opportunity to contest both the factual narrative and the legal claims.

Whatever the outcome, the lawsuit exposes a growing fault line: employers want AI to make organizations more measurable, while workers need assurance that measurement does not turn lawful absence or disability into an invisible penalty.