Lawsuit Claims Meta's AI Tool Metamate Flagged Workers on Leave for Layoffs

Twenty-six former Meta employees allege AI tool Metamate unfairly discriminated against workers on parental or sick leave in layoffs; Meta says humans made final decisions.

Twenty-six former Meta employees have taken the company to federal court in Oakland, alleging that its artificial intelligence systems systematically targeted colleagues who had taken parental leave, maternity leave, or sick leave for dismissal. The suit, filed on July 13 and 14, 2026, challenges the 8,000-job reduction — roughly ten percent of the workforce — that Meta announced this April.

According to the complaint, the social media giant used its in-house AI tool named Metamate to generate termination lists. Data points fed into the system included “token usage” metrics, productivity scores, and even recordings of keyboard strikes and mouse movements. The plaintiffs argue that this algorithmic evaluation disproportionately penalised employees whose absences were protected by law. All 26 individuals had taken some form of statutory leave within the previous 24 months.

One case stands out: a research scientist received her dismissal notice only two days before she was due to give birth.

Meta rejects the accusation outright. The company insists that human managers, not machines, made the final decisions. The AI tools, it says, merely supplied supporting data and were never the sole factor in any termination.

The plaintiffs are asking the court for a temporary restraining order to block the layoffs, which are scheduled to take effect on July 22. They also want an independent audit of the selection process and access to individual arbitration proceedings.

Beyond the individual claims, the case raises far-reaching questions about the use of algorithmic scoring in human-resource decisions. Labour experts warn that such systems risk discriminating against protected groups when they rely on attendance or productivity metrics that penalise legitimate leave.

In Germany, deploying similar technology would face steep legal barriers. Works councils typically must approve any surveillance or performance-evaluation systems under co-determination rules. When making operational dismissals, employers are required by law to weigh criteria such as seniority, age, and dependants — a purely algorithmic ranking would be almost impossible to justify. Additionally, the GDPR places tight restrictions on automated decision-making in employment.

The case, docket number 3:26-cv-07122, is being watched closely as a potential landmark. It may help determine how far companies can go in using AI-driven analytics for major personnel moves.