A scientist on approved pre-birth pregnancy leave was notified of her layoff two days before she gave birth. That is the detail that sticks.
The lawsuit, filed Monday in federal court in the northern district of California, names 26 workers who say Meta used a constellation of internal AI systems to score, rank and select employees for a reduction that cut about 8,000 roles earlier this year. The complaint runs 71 pages. It says the termination list wasn’t built from the considered judgment of managers familiar with the work.
Under the hood, the technical reality was a monitoring program launched quietly this year. It captured keystrokes, mouse activity, browser history, messages, emails and location data on company devices. The rollout came via a low-visibility internal post from an engineer, not a senior leader. On some teams there was no consent prompt and no way to opt out.
The AI models learn from watching really smart people do things
Mark Zuckerberg put it that way in an internal meeting, according to the Information. The average intelligence of the people at this company is significantly higher than the average set of people you can get to do tasks. The program was meant to train Meta’s AI on those behaviors. More than 1,600 employees signed a petition calling it a privacy violation. Zuckerberg paused it in June.
The plaintiffs include an engineer who received a lowered rating because of time off for an injury. A manager on medical leave was let go 16 days into his time off. The complaint argues that productivity metrics do not exist when workers are on medical or family leave, and for people with disabilities those metrics might be reduced. Workers on protected leave ended up overrepresented on the termination list, the complaint argues, because the scoring system not only ignored their leave status but effectively punished them for exercising legal rights.
Meta disputes the allegations. The claims have no merit and no basis in fact, a spokesperson said. Personnel decisions, the spokesperson insisted, are made by humans, not algorithms.
California, Colorado and Illinois have passed laws
They target AI-related bias and automated decision systems in the workplace. The plaintiffs are asking the court for an independent audit of Meta’s AI tools. They are also seeking a preliminary ruling to stop the layoffs from being finalized while the case proceeds. The 26 workers remain Meta employees until 22 July, when their terminations are set to begin.
Their lawyers have asked the judge to allow them to remain anonymous and to preserve their employment status while arbitration is pending. When the separations take effect, the damage cannot be undone: health coverage tied to the employer vanishes during pregnancy, postpartum recovery, and ongoing medical treatment; leave rights with deadlines expire; unvested equity is forfeited; immigration status is jeopardized.
