A legal challenge has emerged against Meta's May redundancy round that eliminated 8,000 positions, representing roughly 10 per cent of the company's workforce. Twenty-six current employees filed suit in federal court in Oakland, California on July 13, claiming the technology giant deployed artificial intelligence systems to identify candidates for dismissal in ways that systematically disadvantaged those absent due to medical, parental or family leave. The plaintiffs contend that Meta's algorithmic approach, which incorporated keystroke monitoring, activity analysis, token-usage tracking and performance scoring, created inherent structural barriers for workers exercising their legal rights to protected leave.
The core dispute centres on how Meta's systems measured and weighted employee contributions without accounting for lawful absences. According to the legal filing, numerous metrics used to evaluate performance "by design, cannot be accumulated by an employee who is on protected medical or family leave, or whose output is reduced by a disability." This architectural flaw meant that workers utilising their statutory entitlements faced systematic disadvantage in the algorithm's calculations. The complaint emphasises that Meta failed to pause these automated systems or conduct the individualised, leave-neutral assessments that employment law demands before making termination decisions.
The demographic composition of the plaintiff group underscores the alleged disparate impact. Approximately half had taken time for caregiving or pregnancy-related reasons, including eight women who took maternity or pregnancy-related leave, four men who took parental leave, and one woman who took leave for family care and bereavement. Each of the 26 anonymous claimants had requested or received workplace accommodations for disabilities or were exercising protected leave rights when notified of their impending redundancy. Though terminations were scheduled to commence on July 22, all remained on Meta's payroll at the time of filing.
One particularly troubling allegation involves a plaintiff who disclosed a serious health condition and disability that Meta's own healthcare provider had approved. The lawsuit claims his manager actively discouraged him from pursuing the necessary medical leave, warning that doing so would trigger selection for the anticipated redundancy round. Rather than providing reasonable accommodations for his condition, Meta allegedly used his reduced output during medical absence as justification for elimination. Such conduct, the plaintiffs argue, violates the spirit and letter of disability rights protections.
Meta has dismissed the allegations as lacking factual foundation, asserting in a statement that "workforce management and organisational decisions were and are made by people, not AI." This defence essentially argues that humans made the final determinations, notwithstanding the algorithmic frameworks that shaped the data those decision-makers reviewed. The company's characterisation downplays the structural role that automated systems played in filtering candidates and generating the performance rankings that informed human choices. Critics argue this framing obscures how algorithmic bias can operate through ostensibly human decision-making processes.
The legal action invokes multiple federal and state employment statutes. The Family and Medical Leave Act provides job-protected status for qualifying absences, whilst the Americans with Disabilities Act mandates reasonable accommodations for disabled workers. The Pregnancy Discrimination Act and the newly enacted Pregnant Workers Fairness Act extend protections to pregnant workers and those with pregnancy-related conditions. Collectively, these laws create a framework designed to prevent employers from penalising workers for exercising protected rights. The plaintiffs contend Meta's system violated all four statutory regimes.
Crucially, the complaint relies on "disparate impact" liability, a civil rights doctrine establishing that facially neutral policies may constitute discrimination if they disproportionately burden protected classes without legitimate job-related justification. Whilst this principle has underpinned employment law for decades, the Trump administration has recently signalled hostility to disparate impact enforcement, instructing federal agencies to deprioritise such cases on grounds that the doctrine undermines meritocracy. The administration's stance has already influenced enforcement priorities, with the Equal Employment Opportunity Commission dropping certain complaints.
Yet the Meta litigation demonstrates that companies cannot simply rely on executive branch reluctance to prosecute disparate impact claims. Employees retain the right to pursue such litigation independently if federal authorities decline involvement, particularly when state law provides overlapping protections. Several states have enacted statutes specifically prohibiting disparate impact discrimination, insulating workers in those jurisdictions from federal policy shifts. California's robust employment protections position these plaintiffs to advance their claims despite the broader political environment.
The plaintiffs' legal team emphasises that algorithmic performance tracking creates particular vulnerability for women, who disproportionately utilise pregnancy and caregiving leave. By recording such absences as reduced performance without contextual adjustment, Meta's system mathematically disadvantaged a protected group. This statistical disparity forms the foundation of disparate impact claims under Title VII of the Civil Rights Act, relying on a landmark 1971 Supreme Court precedent recognising the doctrine's validity. The lawyers argue the case illustrates how AI-enabled discrimination can operate at scale and velocity that traditional hiring discrimination could not achieve.
The plaintiffs seek to preserve the employment status quo pending arbitration, recognising that finalised separations create irreversible harms. Loss of employer-subsidised health coverage during pregnancy and postpartum recovery, forfeiture of time-bound leave rights, destruction of unvested equity compensation, and triggered immigration consequences cannot be restored once terminations become effective. This urgency reflects the disproportionate vulnerability of workers on protected leave, whose financial and legal circumstances make severance particularly consequential. The case raises fundamental questions about whether companies deploying AI systems bear heightened obligations to ensure those systems do not perpetuate discrimination against legally protected groups.
