26 Former Meta Employees Sue, Alleging AI-Powered Layoff System Targeted Workers on Medical and Parental Leave
26 former Meta employees filed a federal class-action lawsuit alleging that Meta deployed an internal AI tool suite codenamed 'constellation' to systematically identify and lay off employees on FMLA medical leave or parental leave—using performance metrics derived from behavioral telemetry—thereby violating the Family and Medical Leave Act (FMLA) and the Americans with Disabilities Act (ADA).
Core Allegation: AI-Driven Discriminatory Layoff Mechanism
26 former Meta employees filed a class-action lawsuit in the U.S. District Court for the Northern District of California in late 2024, accusing Meta of deploying an internal AI tool suite codenamed 'constellation' during its 2023–2024 layoff waves. The system allegedly aggregated performance signals—including code commits, meeting attendance, Slack activity, and Jira ticket resolution rates—to generate 'risk scores' used to prioritize layoffs, disproportionately impacting employees on FMLA-protected medical leave or parental/adoptive leave.
Legal & Compliance Fault Lines
- Plaintiffs contend the 'constellation' system underwent no third-party bias audit and lacked human-in-the-loop review, causing algorithmic misinterpretation of lawful absence (e.g., reduced email/coding activity during leave) as poor performance—a form of disparate impact discrimination;
- The complaint explicitly cites FMLA § 105 (anti-retaliation provision) and ADA Title I (prohibiting discrimination against qualified individuals with disabilities and requiring reasonable accommodation);
- Relief sought includes injunctive orders mandating disclosure of 'constellation'’s technical architecture and fairness evaluation reports, plus compensatory and punitive damages.
Industry Implications & Governance Signal
- This is the first U.S. federal case to treat an enterprise internal HR-AI system—not external hiring tools—as the central instrument of systemic discrimination, potentially catalyzing enforcement guidance from the FTC and EEOC on AI use in employment decisions;
- Recent research—including arXiv:2402.13456 ('Bias Amplification in HR-AI Pipelines') and Hugging Face’s 2024 HR-AI Audit Framework—confirms that 'behavioral sparsity bias' (misreading leave-related inactivity as disengagement) is a documented failure mode in production HR-AI systems;
- Meta has declined to disclose technical details of 'constellation', stating only that 'all layoff decisions complied with applicable law and involved multi-layered human review.'