A group of 26 Meta employees has commenced legal action against the technology giant, alleging that the company deployed artificial intelligence systems to identify and select workers for redundancy, with a particular focus on those taking medical, parental, or family leave. The lawsuit, filed in federal court in Oakland, California on July 13, represents part of Meta's broader workforce reduction announced in May, when the company indicated it would eliminate approximately 8,000 positions—roughly 10 per cent of its total workforce. The plaintiffs argue that the company's reliance on algorithmic performance assessments, keystroke monitoring, activity tracking, AI token-usage metrics and automated ranking systems created an inherently discriminatory selection process that systematically disadvantaged employees unable to maintain productivity during protected absences.
According to the complaint, Meta's selection methodology fundamentally failed to account for the reality that workers on protected leave cannot accumulate performance scores or maintain the same output levels as colleagues working full hours. The lawsuit states that the company's systems were designed in ways that rendered it impossible for employees on qualifying medical or family leave to generate the metrics and ratings that would protect them from redundancy. Critically, the plaintiffs contend that Meta made no effort to pause these algorithmic assessments during the review period to conduct the individualised, leave-neutral evaluation that employment law requires. This architectural flaw in the company's decision-making process meant that absence itself became a proxy for poor performance, regardless of the quality or contribution of employees' actual work.
The 26 anonymous plaintiffs, all of whom remain employed pending separations scheduled to begin July 22, each took protected leave and sought or received workplace accommodations for disabilities. Approximately half had taken leave related to pregnancy, childbirth, or caregiving responsibilities. Eight women had taken maternity or pregnancy-related leave, four men had accessed paternal leave, and one woman had taken leave to care for a family member before subsequently requesting bereavement leave. Beyond these statistics lies a pattern that the lawsuit argues reflects systemic discrimination: one male plaintiff disclosed that his manager actively discouraged him from taking approved medical leave for a serious health condition and disability, explicitly warning that doing so would result in his selection for the anticipated layoffs. The manager, according to the filing, offered no accommodation for the disability itself, creating a coercive environment that pressured the employee to forgo legally protected rights.
Meta has rejected the allegations, issuing a statement asserting that the claims "lack merit and are not based on facts," and emphasising that "workforce management and organisational decisions were and are made by people, not AI." The company's response mirrors a broader corporate narrative in which AI systems are presented as tools that inform human judgment rather than autonomous decision-makers. However, the lawsuit presents a more complex picture: the plaintiffs' legal team argues that even if human managers made final determinations, those decisions flowed from an algorithmic process that systematically embedded discrimination at its foundation. In this framing, the question is not whether humans made the ultimate call, but whether the data fed into that human decision-making was itself tainted by a system incapable of distinguishing between reduced output caused by protected absence and genuine performance deficiency.
The legal claims traversed multiple federal and state statutes, including the Family and Medical Leave Act, the Americans with Disabilities Act, the Pregnancy Discrimination Act, and the Pregnant Workers Fairness Act. Notably, the lawsuit also invokes the concept of "disparate impact liability," a civil rights doctrine holding that ostensibly neutral policies can constitute unlawful discrimination if they disproportionately burden protected classes of workers, regardless of whether discriminatory intent is proven. This framework has become increasingly contentious in recent months. The Trump administration has directed federal agencies to deprioritise disparate impact enforcement, contending that the doctrine undermines meritocratic principles and falsely assumes that workforce imbalances automatically signal discrimination. These policy shifts have already influenced enforcement actions; the Equal Employment Opportunity Commission has discontinued discrimination cases on behalf of some workers in alignment with the administration's position.
Yet the Meta lawsuit demonstrates that despite federal retreat from disparate impact enforcement, private litigation remains viable and consequential. Workers retain the capacity to pursue such claims independently through the courts, and several states maintain statutory prohibitions against disparate impact discrimination that operate independently of federal policy shifts. California, where the lawsuit was filed, explicitly recognises disparate impact liability in its employment discrimination statutes, creating a jurisdictional safe harbour for this legal theory. The plaintiffs' counsel has argued specifically that Meta's "algorithmically assisted selection process" systematically recorded absences as reduced performance, thereby placing disproportionate burden on women—who account for the vast majority of pregnancy and primary caregiving leave-takers across most industries and organisations.
The litigation carries particular significance for Malaysia and Southeast Asia as technology companies in the region increasingly adopt similar AI-driven workforce management systems and performance analytics platforms. Meta's operational footprint extends substantially across the region, and the company's approach to algorithmic decision-making influences how multinational technology firms and their regional competitors structure human resources functions. Local regulatory frameworks in Malaysia, including the Employment Act and ongoing developments in data protection law, may create similar vulnerabilities to those exposed in the California lawsuit. Malaysian workers and their advocates should pay attention to how courts resolve questions about algorithmic opacity, the burden of proof regarding discriminatory intent, and whether algorithmic systems that produce discriminatory outcomes qualify as violations of local employment protection statutes.
The immediate relief sought by the plaintiffs is relatively modest in scope but profound in implication: they ask the court to preserve the status quo, maintaining the 26 workers in employment pending arbitration of their claims. The legal team emphasised that once layoffs become final, the harms become irreversible—employer-subsidised health coverage lapses during critical periods of pregnancy, postpartum recovery, and active medical treatment; time-limited leave entitlements expire unused; unvested equity forfeits entirely; and immigration sponsorships, particularly relevant for a technology workforce that includes substantial numbers of skilled migrants, may trigger visa complications. These irreversibilities underscore why employment discrimination claims involving leave and disability deserve urgent attention before separations take effect.
The dispute also illuminates a broader tension in how society regulates artificial intelligence systems deployed in consequential contexts. Meta's insistence that people, not algorithms, made decisions reflects a common corporate positioning but sidesteps the substantive question of whether human decision-makers were operating with corrupted inputs. If the data provided to human managers systematically underrepresented employees on protected leave, human judgment cannot cure the upstream discrimination embedded in the system architecture. For Malaysian employers and policymakers considering implementation of similar algorithmic performance and workforce management tools, the case underscores the necessity of conducting disparate impact assessments before deploying AI systems that will affect employment decisions—particularly regarding leave, disability accommodations, and other legally protected statuses where systemic disadvantage risk exists.
The case arrives at a moment when both the regulatory environment and technological capability for AI-driven employment decisions are evolving rapidly. Federal enforcement of discrimination law is loosening in the United States under current administration priorities, yet courts retain authority to apply existing statutory frameworks. Technology companies face growing employee mobilisation around algorithmic accountability, particularly concerning systems that lack transparency and create concentrated power to determine livelihoods. For Meta specifically, the litigation follows previous disputes over algorithmic systems and workforce management, suggesting a pattern of legal contestation around the company's approach to employment automation. Whether the court grants the plaintiffs' emergency request to preserve their employment status pending arbitration will provide early signals about judicial receptivity to disparate impact claims in the AI employment context.
