Meta's Metamate AI Under Fire: 26 Employees Sue Over Biased Layoff Algorithm

The lawsuit against Meta exposes a critical flaw in enterprise AI: when algorithms trained on productivity data are used to rank employees, they can systematically disadvantage those with legitimate absences. The allegation that Metamate and a keystroke-based productivity score biased the layoff selection toward able-bodied workers highlights the need for rigorous bias testing in AI systems before they influence life-altering decisions. For AI developers and researchers, this case is a real-world test of algorithmic accountability.

Meta is facing a first-of-its-kind legal challenge after 26 current and former employees filed a lawsuit in federal court in Oakland, California, accusing the company of using artificial intelligence to deliberately target workers with medical conditions for its May 2026 layoffs. The plaintiffs, proceeding anonymously, allege that Meta’s AI-driven evaluation systems—including an internal large language model assistant called Metamate, a workplace communication monitoring tool dubbed the 'second brain,' and a productivity score derived from keystrokes and screen activity—systematically penalized employees who had taken medical, disability, or family leave. According to the complaint, these AI tools factored in metrics such as AI token usage and output volume that, by design, could not be accumulated by someone on leave or with a disability, effectively automating a discriminatory filter. The layoffs, which eliminated roughly 8,000 positions (10% of Meta’s global workforce), are scheduled to take effect on July 22, 2026, and the lawsuit seeks a temporary halt while the claims proceed through individual arbitration.

“

The layoffs, which eliminated roughly 8,000 positions (10% of Meta’s global workforce), are scheduled to take effect on July 22, 2026, and the lawsuit seeks a temporary halt while the claims proceed through individual arbitration.

The legal action arrives amid a broader reckoning over algorithmic bias in employment. In 2023, Workday was sued over claims that its AI screening tools discriminated against job applicants based on race, gender, age, and disability—a case still ongoing. California’s AB 2930, proposed in 2024, aimed to ban 'algorithmic discrimination' in hiring and firing, but died in the legislature after fierce opposition from tech industry groups. New York City and California now have bias-testing requirements for automated employment decision tools, but enforcement is nascent. The Meta lawsuit could become a landmark test of whether these laws—and existing federal anti-discrimination statutes like the Americans with Disabilities Act—can hold companies accountable when AI systems produce disparate impacts on protected groups.

Meta has vigorously denied the allegations. A company spokesperson stated, 'Workforce management and organizational decisions were and are made by people, not AI,' and called the claims without merit. This defense—that human managers ultimately signed off on the termination lists—mirrors a common industry argument that AI merely provides recommendations, not final decisions. However, the plaintiffs contend that the AI-generated rankings were determinative, with managers rubber-stamping lists produced by algorithms that had already disadvantaged certain employees. The distinction will be crucial in court, as it could define the legal liability of employers who rely heavily on opaque, data-driven ranking systems for headcount decisions.

From an industry perspective, the case exposes a deep tension: many large tech companies, including Meta, are significantly increasing investments in AI for internal productivity tools while simultaneously reducing headcount. Meta CEO Mark Zuckerberg has stated that he expects no further company-wide layoffs in 2026, but the restructuring is part of a shift toward greater automation. The lawsuit alleges that this cost-cutting drive created incentives to use AI in ways that circumvented human oversight and legal protections. If successful, the suit could compel companies to reevaluate how they design, audit, and govern AI systems that influence employment outcomes, pushing for more transparency and human-in-the-loop safeguards.

For the labor market, the case underscores the risks of relying on quantified productivity metrics that fail to account for legitimate absences. The plaintiffs’ argument that AI usage as a metric is inherently biased against those on leave highlights a design flaw that could affect any workplace that tracks digital activity. As AI becomes more embedded in performance management, HR departments will need to ensure that algorithms are tested for adverse impact and that accommodations are built into the system, not left to after-the-fact human intervention.