Meta faces lawsuit over AI-driven layoffs targeting workers with disabilities

In the spring of 2026, Meta announced it would cut ten percent of its workforce—nearly eight thousand people—in a series of layoffs beginning in May. The company framed the decision as a necessary restructuring. But twenty-six former employees, now scattered across six states and Washington, D.C., believe something else happened: that the company used artificial intelligence to systematically identify and remove workers with disabilities or those who had taken medical leave.

The lawsuit, filed in federal court in Oakland, California on July 13, alleges that Meta's AI system evaluated workers using metrics like productivity scores and AI token usage—a measure of computational resources consumed. For employees who had missed work due to medical conditions, these numbers looked worse on paper. The algorithm, the plaintiffs argue, treated absence as a liability and flagged those workers for termination. What appeared to be an objective, data-driven decision was actually a mechanism that discriminated against some of the company's most vulnerable employees.

The twenty-six plaintiffs are suing under federal and state laws designed to protect workers from exactly this kind of harm: statutes that prohibit discrimination based on disability, that shield employees who take medical leave, and that guard against retaliation for pregnancy-related absences. These are not new protections. They exist because employers have always found ways to rationalize firing people for reasons that have nothing to do with their actual ability to do the job. What is new is the scale and the opacity. An algorithm can make the same biased decision ten thousand times before anyone notices.

Meta's response came quickly. A company spokesman said on July 14 that the claims have no merit, and made a striking assertion: that workforce decisions were made by people, not by artificial intelligence. The statement attempts to draw a bright line between human judgment and algorithmic recommendation. But that line may not hold up under scrutiny. The question at the heart of this case is not whether Meta executives consciously decided to fire disabled workers. It is whether they built a system that did so anyway, and whether they knew—or should have known—what that system would do.

The layoffs themselves were real and substantial. Eight thousand people lost their jobs. But the lawsuit suggests that the way those eight thousand were chosen was not random, not merit-based, and not blind to disability status or medical history. Instead, it was filtered through an AI system that converted human vulnerability into a data point, and that data point into a termination notice.

What happens next will likely turn on discovery—the process by which both sides examine each other's documents and testimony. Meta will have to produce the code, the training data, the metrics, and the decision logs. The company will have to explain how the AI system worked, what it optimized for, and whether anyone tested it for bias before deploying it to make decisions that would upend people's lives. The plaintiffs will have to prove not just that the system was biased, but that Meta knew or should have known it was biased and used it anyway.

The case sits at the intersection of two urgent questions in technology law: whether AI systems can be held accountable for discrimination, and whether companies can hide behind claims of automation when those systems produce discriminatory results. Meta's assertion that humans made the decisions, not machines, may turn out to be the most important statement in the lawsuit—not because it is true, but because it will force the company to explain exactly what role the AI played, and why.