The Scorecard Nobody Admits to Building
Even now, dear reader, in a year when "the algorithm did it" has become corporate America's favorite shrug, Meta wants you to know that no algorithm did anything. Twenty-six current and former employees, filing across California, Florida, Illinois, New York, Pennsylvania, Washington state, and Washington D.C., sued the company on July 13th claiming otherwise: that Meta used internal tools, including its "Metamate" AI assistant, keystroke monitoring, activity dashboards, and AI-token usage trackers, to score and rank employees for a round of roughly 8,000 layoffs, no manager judgment required.
The Machine does not take maternity leave. It also, apparently, cannot conceive of anyone else doing so either.
Structurally Incapable of Seeing You
In a development that will surprise no one who has been paying attention, the scoring system's blind spot lines up precisely with the workers who could least afford one. The complaint alleges the tools measured output and activity, full stop, which meant employees on medical or family leave accumulated none of it while they were, correctly, not working. The filing's own language is the kicker here: the system was "structurally incapable of capturing employees on protected leave." Not biased against them, exactly. Just unable to register that they existed.
Plaintiffs allege this amounts to discrimination against workers with disabilities and pregnancies under federal and state law, and that Meta skipped the bias testing California and New York City both require before deploying tools like this on a workforce.
One presumes a dashboard that cannot see you is, from the dashboard's perspective, a fairly efficient way to forget you.
The Judge Wants to See the Room
Sources confirm the plaintiffs asked U.S. District Judge William Orrick for a temporary restraining order to halt the layoffs, scheduled for July 22nd, while arbitration proceeded. Orrick declined, ruling the plaintiffs "were not in the rooms where it happened" and therefore lacked evidence that AI, rather than a human manager, made the call. Meta's own position is that "workforce management and organisational decisions were and are made by people, not AI" — an assertion the company has made in a lawsuit specifically about whether that's true.
Sources within the Legal Community note the deeper problem: most of the plaintiffs are bound by arbitration agreements, which keeps their cases out of open court and off each other's radar entirely. As attorney Christine Webber put it, "Even if you establish that a particular system would produce discriminatory outcomes left and right, you have no way of sharing that information with other employees." A preliminary injunction hearing is set for August 24th. The layoffs already happened.
Where This Leaves Us
And yet — as if this were not enough — the actual dispute here isn't really about whether Metamate typed the pink slips. It's about whether a company can build a measurement system that only sees people who are, at that exact moment, producing, and then act surprised when it discriminates against everyone who briefly wasn't. That is not a bug report. That is a design document.
The algorithm was unavailable for comment. It was also, structurally, unavailable for everyone on leave.
Sources: IBTimes UK — 'AI Decided Who Got Fired': Meta Sued Over Claims AI Selected Workers on Maternity and Medical Leave for Layoffs · Allwork.Space — Meta Layoff Case Shows Why Workplace AI Discrimination Is So Difficult To Prove
