A Judge Wouldn't Stop Meta's Layoffs. He Also Said the AI Discrimination Claims Raise "Serious Questions."

Twenty-six Meta employees went to federal court in Oakland last week arguing something relatively new for a layoff case: that an algorithm, not a person, decided they should lose their jobs - and that the algorithm could not tell the difference between someone who was not performing and someone who was on approved medical leave. On July 17, a judge agreed the question was serious enough to take seriously. He just would not stop the layoffs to answer it.

The case grows out of the roughly 8,000 job cuts - about 10% of Meta's global workforce - that the company announced in May as part of its push to redirect spending toward AI infrastructure, the same round AINews covered on July 15 when Mark Zuckerberg conceded the AI investment was not yet paying off. What was not clear at the time was exactly how Meta decided which employees, among those thousands, would go.

What the lawsuit says

According to the lawsuit, filed July 13 in the U.S. District Court for the Northern District of California, Meta ranked employees using a mix of internal AI tools - including its Metamate assistant - alongside keystroke and click-tracking software, AI token-usage dashboards, and algorithmically generated performance scores. That tracking traces back to an internal program the suit calls the Model Capability Initiative, which it says Meta installed on U.S. employees' work laptops starting in April with no opt-out, logging keystrokes, mouse movement, and periodic screenshots.

The plaintiffs - among them employees on medical, parental, pregnancy, disability, and other protected leave or accommodations, according to filings and wire coverage - say the scoring system had no mechanism to account for approved time away from a keyboard. Weeks or months of legally protected leave simply registered as missing data, the complaint alleges, dragging down aggregate scores with no leave-neutral adjustment applied before rankings were finalized.

They are suing under the Family and Medical Leave Act, the Americans with Disabilities Act, the Pregnancy Discrimination Act, the Pregnant Workers Fairness Act, and Title VII's disparate-impact provision - a combination that, if it holds up, would make this one of the first real legal tests of whether AI-assisted layoff selection can itself amount to discrimination.

Why Orrick let the layoffs proceed

U.S. District Judge William Orrick declined to grant the emergency relief the workers wanted: a temporary restraining order preserving their jobs while the underlying claims are argued in arbitration. Orrick found the plaintiffs had not shown the "irreparable harm" that emergency injunctions require - a familiar, high bar in employment law, where lost wages are usually treated as recoverable rather than irreversible. The layoffs went forward as scheduled on July 22.

But Orrick did not wave the underlying claims away. According to plaintiffs' attorneys, his written order stated the case raises serious questions going to the merits, and that he could revisit his decision if the parties produce more evidence on whether and how AI actually drove the termination decisions. A separate motion for a longer-term preliminary injunction remains pending. Meta, for its part, maintains there is nothing to revisit: "Workforce management and organizational decisions were and are made by people, not AI," a company spokesperson said, calling the claims meritless.

The question arbitration has to answer

That is the crux the case will have to settle in arbitration - not whether Meta used AI tools somewhere in its process, because plenty of large employers now do in one form or another, but whether the specific outputs of those tools drove specific termination decisions, and whether anyone checked those outputs against the legal protections some affected employees were entitled to before finalizing them.

That is a narrower and more provable question than "did an algorithm do this," and employment lawyers watching the case say it is the kind of question that will define the next wave of AI-in-HR litigation regardless of how this particular dispute resolves. For a company that spent its May earnings call explaining why AI spending would eventually pay off, its lawyers are now spending July explaining, in a different courtroom, exactly how little of that AI was supposedly involved in one of its highest-stakes people decisions of the year.

Sources

Reuters via Investing.com, US judge won't block Meta from laying off workers who filed AI discrimination lawsuit: https://www.investing.com/news/stock-market-news/us-judge-wont-block-meta-from-laying-off-workers-who-filed-ai-discrimination-lawsuit-4798876

Associated Press via ABC News, 26 Meta employees sue, alleging layoff picks hit workers on medical, parental leave: https://abcnews.com/Technology/wireStory/26-meta-employees-sue-alleging-ai-driven-layoff-134764788

SFGATE, Meta laid off thousands to prioritize AI. Former employees say AI was used to lay them off: https://www.sfgate.com/tech/article/meta-disability-lawsuit-22347135.php

Complaint, Doe et al. v. Meta Platforms, Inc., N.D. Cal., filed July 13, 2026: https://www.courthousenews.com/wp-content/uploads/2026/07/meta-employee-complaint.pdf