26 Employees Sue Meta Over AI-Powered Layoff Discrimination
Alleged AI systems scored workers on keystrokes and token usage, penalizing those on leave.
A group of 26 anonymous Meta employees has filed a lawsuit alleging the company used discriminatory AI-powered systems during its May 2025 layoffs, which affected 8,000 workers (10% of the workforce). According to the complaint filed in the Northern District Court of California, Meta relied on a "constellation" of internal AI tools—including Metamate, an internal large-language model assistant, and algorithmic productivity scores based on keystrokes, browser history, and email data—to rank and select employees for termination. The systems also leveraged AI token consumption records, which the plaintiffs argue penalized workers for reduced output during protected leaves.
Specific examples include a scientist two days before giving birth and a manager on pregnancy-related disability leave who were flagged for layoff. The lawsuit claims Meta was aware of the bias but failed to pause the system. Plaintiffs are seeking a preliminary injunction to block the layoffs scheduled for July 22 while they pursue private arbitration. Meta denies the allegations, stating layoff decisions were made by people, not AI. This comes just months after a separate discrimination suit over age bias in February 2025 layoffs.
- 26 anonymous employees allege Meta used Metamate (internal LLM), keystroke metrics, and AI token consumption to score and rank workers for layoffs.
- The systems allegedly penalized workers on protected leave (maternity, disability), leading to disproportionate selection; one scientist was two days from giving birth.
- Plaintiffs seek a preliminary injunction to block July 22 layoffs; Meta denies claims, insisting decisions were human-made.
Why It Matters
This case highlights the risks of using AI in HR decisions, especially bias against protected leave-taking employees.