Meta AI-driven layoff discrimination lawsuit

The structural flaw at the center of this lawsuit is not that Meta's AI was programmed to target people on leave - it is that the metrics chosen made leave mathematically indistinguishable from disengagement. According to the complaint, Meta's layoff selection drew on keystroke activity, browser history, AI token-usage dashboards, and ratings for being 'AI-native' - all measures that, by definition, require physical presence at a keyboard to accumulate. ^[1]An employee on parental or medical leave cannot log keystrokes, consume AI tokens, or rack up 'AI-native' ratings during their absence. The system, as alleged, 'failed to distinguish employees' leave from lack of engagement,' and at least one director was warned that absences could be 'held against him' during the layoff evaluation. ^[2]This is the crux of the disparate impact argument: no explicit intent to discriminate is required. A facially neutral productivity system that cannot accommodate absence will statistically burden any class of workers who take protected leave - women on maternity leave, workers with disabilities, employees recovering from illness. The complaint states the AI scoring metrics 'by design, cannot be accumulated by an employee who is on protected medical or family leave, or whose output is reduced by a disability.' ^[3]Critically, the lawsuit also alleges Meta 'did not pause the system for the individualized, leave- and accommodation-neutral review that the law requires,' and that some plaintiffs were actively discouraged from taking FMLA leave. ^[4]The result is a system where the path of least resistance - staying on leave as entitled by law - became a career liability.

The plaintiffs' legal theory rests almost entirely on disparate impact doctrine - the civil rights principle that facially neutral policies can constitute illegal discrimination if they disproportionately burden a protected class, even without discriminatory intent. ^[5]That doctrine is now under active political threat: the Trump administration has moved to abandon disparate impact liability at the federal level, which means the case's viability may hinge on whether the courts independently uphold the doctrine in the current legal environment. ^[5]Meta's defense - that workforce decisions were made by people, not AI - invokes a common corporate shield, but employment law analysts push back on its sufficiency. The argument, as summarized in coverage, is that 'a human approved it does not constitute sufficient oversight when the upstream scoring methodology was built without legal-leave exemptions.' ^[1]In other words, a manager rubber-stamping an algorithm's ranked output does not reset the discrimination clock if the ranking criteria were themselves discriminatory. The EEOC has issued guidance affirming that AI tools in employment decisions remain subject to the ADA and Title VII regardless of whether a human is nominally in the loop. ^[1]Compounding the urgency: terminations are scheduled to begin July 22, 2026 - just eight days after the suit was filed - which is why plaintiffs are seeking a preliminary injunction. Their counsel argues that once separations are finalized, harms including health coverage loss during active medical treatment, forfeiture of unvested equity, and adverse immigration consequences become irreversible. ^[3]

This lawsuit does not target an outlier practice. Roughly 87% of companies use AI in HR decisions, meaning the question of whether leave-agnostic productivity metrics constitute unlawful disparate impact reaches well beyond Meta. ^[6]The Meta case is, however, the first major federal lawsuit to target AI-driven layoff selection specifically - earlier cases focused on hiring (Amazon's abandoned resume-screening tool, the EEOC's settlement with iTutorGroup over age-discriminatory hiring algorithms, and the Workday lawsuit alleging bias in recruiting software). ^[7]A ruling against Meta - particularly one that establishes that human sign-off on AI output is insufficient oversight - could force a structural redesign of performance management tools across the industry, requiring systems to flag, exclude, or separately calibrate employees on protected leave before generating ranked lists. California's automated decision-making rules add a state-level dimension that could independently constrain how companies operating there deploy such systems. ^[8]On social media, the lawsuit generated substantial public reaction, with posts framing the case as AI targeting workers with medical conditions drawing significant engagement and amplifying the human stakes of algorithmic HR. The reputational pressure compounds Meta's legal exposure: this is the second major discrimination lawsuit tied to its workforce reductions in 18 months, following a February 2025 age discrimination suit over prior layoff rounds. ^[9]