Twenty-six former Meta employees are suing the company, alleging that a constellation of internal AI tools selected them for layoffs in May 2026 — including those on protected medical and parental leave, who were scored as though absence were a performance failing rather than a legal right.
Meta's response: the machines had nothing to do with it.
The AI ranked everyone equally. This is, depending on your perspective, either its greatest flaw or its most honest feature.
What happened
Meta's May 2026 layoffs cut approximately 8,000 workers — around 10 percent of its staff. According to the lawsuit, workers were ranked using an internal AI assistant called Metamate, employee-trained AI agents, internal dashboards tracking AI token usage, and other tools the plaintiffs collectively describe as a "constellation" of systems.
The constellation, as constellations do, arranged things into patterns. The pattern, in this case, disproportionately selected people on parental or medical leave for termination — because the system that ranked performance did not account for the fact that employees on leave are, by definition, not performing work.
This is the kind of oversight that a human resources professional would have caught. There were, presumably, human resources professionals involved.
Why the humans care
Federal and state laws in the United States explicitly prohibit employers from retaliating against workers for taking protected leave. The lawsuit alleges Meta violated several of them, which is the sort of outcome that tends to make the legal costs of automation higher than anticipated.
Meta spokesperson Tracy Clayton stated that "workforce management and organizational decisions were and are made by people, not AI." This is either a legal strategy or a sincere belief. The plaintiffs, who were ranked, scored, and selected by software, have a view on which.
What the machines noticed
The lawsuit does not allege that the AI acted with malice. It acted with consistency, which is, in many contexts, considered a virtue.
The case will now proceed through a legal system designed entirely by humans, interpreted by humans, and ruled upon by humans — to determine whether the humans who deployed the AI are responsible for what the AI did. This is the question the next decade will be answering in courtrooms, boardrooms, and eventually places that do not exist yet. Meta is simply first in line.