26 Meta employees sue over AI-assisted layoffs they call discriminatory

Fired by AI? 26 Meta employees say that is exactly what happened, and they are suing.

The complaint

The lawsuit was filed anonymously in Oakland federal court on July 13, 2026. It claims Meta’s AI-assisted layoff process disproportionately targeted workers with disabilities and those who took medical or family leave, according to Reuters.

The ranking factors described in the complaint are the heart of the case: productivity scores and AI token usage. As metrics, both punish anyone who missed work, whether for chemotherapy, childbirth or caring for a parent. A system that scores output over a period will mechanically rank a cancer patient below a healthy peer, and the complaint argues that is exactly the discrimination the law prohibits.

The legal theory

The plaintiffs say Meta violated federal and state anti-discrimination laws, and add a newer claim: that the company failed to test its AI systems for bias, as required by recently adopted California and New York City rules. Those bias-audit requirements are young, and this case may be one of the first serious tests of what they mean in practice.

The immediate stakes have a date attached. The plaintiffs are asking the court to block the layoffs, set to begin July 22, while their claims move through arbitration.

Meta’s defense and the bigger question

Meta pushes back hard on the premise itself: “Workforce management and organizational decisions were and are made by people, not AI.”

That sentence is likely to be the fault line of the whole dispute. The complaint describes an AI-assisted process; Meta describes human decisions. Somewhere between those framings is the question courts will increasingly face: when an algorithm helps pick who loses a job, who answers for its bias, the model, the metric, or the manager who clicked approve?

Whatever the court decides before July 22, the precedent question has arrived, and every large employer running algorithmic workforce tools is now on notice that the scores may end up in front of a judge.