Meta's AI layoffs hit 26 employees on leave, sparking bias suit

For AI and machine learning professionals, this lawsuit is a textbook case of how biased training signals—using keystrokes, token usage, and activity data—can produce systematically adverse outcomes. When an algorithm is designed to reward volume of digital output, employees on leave unavoidably get penalized, transforming a neutral metric into a legally actionable fairness problem. As organizations scramble to integrate AI into people analytics, the need for built-in fairness constraints and human oversight has never been more urgent.

A group of 26 Meta employees has filed a federal lawsuit in Oakland, California, alleging that the company’s use of artificial intelligence tools to guide its May 2026 layoffs unlawfully discriminated against workers on protected medical, parental, or family leave. The lawsuit, filed on July 17, 2026, charges that Meta relied on internal AI systems—including keystroke and activity monitoring, AI token-usage dashboards, and algorithmically assisted performance rankings—to select employees for separation among the roughly 8,000 workers, or about 10% of the company’s workforce, being let go. All 26 plaintiffs remain employed for now, with their separations scheduled to begin on July 22, 2026.

The core legal claim is that Meta’s automated performance-evaluation infrastructure systematically disadvantages employees who are on leave. Because metrics such as keystrokes, code commits, or token usage cannot be accumulated during protected absences, the lawsuit argues that the algorithms "by design" produce lower scores for those individuals, leading to their disproportionate selection for layoffs. The complaint further alleges that Meta did not pause the system for the individualized, leave-neutral review required by law, and that one employee was actively discouraged by a manager from taking approved medical leave because doing so would increase the risk of losing their job. About half of the plaintiffs had taken caregiving or pregnancy leave, and the group includes eight women.

Meta has denied the allegations, stating that "workforce management and organizational decisions were and are made by people, not AI." The company’s position places it at the center of a growing debate over the role of algorithmic decision-making in human resources. As enterprise adoption of AI-driven hiring, performance management, and workforce planning tools accelerates, this case tests the legal boundaries of using such systems when they intersect with protected characteristics and rights under the Family and Medical Leave Act, the Americans with Disabilities Act, and similar state laws.

The implications for the broader AI industry are significant. If the plaintiffs prevail, the case could establish that employers using AI-based performance ranking tools bear a duty to calibrate those systems to avoid disparate impact on employees taking legally protected leave. This would force technology companies and other large employers to conduct rigorous algorithmic audits, implement fairness constraints, and maintain human-in-the-loop oversight over high-stakes personnel decisions. Conversely, a ruling in Meta’s favor might embolden more aggressive deployment of productivity-tracking AI without special accommodations for leave-related data gaps.