Meta accused of using biased AI targeting for mass layoffs

A group of 26 former Meta employees is suing the company, claiming its AI-powered performance ranking system effectively punished them for taking legally protected leave, according to Reuters. The lawsuit alleges that Meta's internal AI tools kept scoring employees during parental and medical leave, treating their absence as poor performance rather than protected time off.

This isn't just about one company's layoff process. It highlights a broader problem as more organizations use AI to make workforce decisions at scale. When AI systems rank performance without understanding context like protected leave, they can systematically disadvantage people exercising basic labor rights. The algorithms don't know the difference between someone checked out and someone on maternity leave.

Meta ran multiple rounds of layoffs starting in 2022, cutting over 20,000 jobs. The plaintiffs argue the company relied on a "constellation" of AI monitoring tools that tracked productivity metrics constantly. If you were on leave, you simply weren't producing, and the system flagged you accordingly. The result was what labor lawyers call disparate impact, where a neutral-sounding policy (rank by output) creates discriminatory outcomes (penalizing new parents and people with medical needs).

The case puts AI accountability front and center. Companies love automated decision-making because it scales, but these systems inherit the blind spots of whoever built them. If your training data doesn't flag protected leave as a special case, your model will treat it like any other gap in activity. That's a design choice, not an inevitability.

This matters beyond Meta. Performance management AI is spreading across industries, and most workers have no visibility into how they're being scored. You might think you're doing fine, but an algorithm could be docking you for taking your kid to the doctor or recovering from surgery. The legal system is now being asked to decide whether companies can outsource these decisions to black-box tools.

The lawsuit will likely focus on whether Meta's AI tools violated the Family and Medical Leave Act and similar protections. But the broader question is whether we're comfortable letting algorithms make high-stakes decisions about people's livelihoods without meaningful human oversight or transparency.

What this means for you: If you manage people or build HR tools, audit your performance systems for protected categories. A simple AI workflow: "Review our performance tracking data for the last 12 months. Flag any patterns where employees on documented parental leave, medical leave, or disability accommodation scored lower than their pre-leave baseline. Calculate whether leave-takers were terminated at higher rates than the general population." If you're an employee, document your performance before taking protected leave and know your rights, because the AI evaluating you might not.