When companies discuss integrating artificial intelligence into the workplace, they typically promise efficiency, speed, and enhanced decision-making. But what happens when the system evaluates an employee who did not type, write code, or utilize AI tools, simply because they were away on parental or sick leave?

This is the central question in a new lawsuit filed by 26 employees against Meta. They allege that the tech giant utilized AI-based systems and activity metrics to rank employees and determine who would be laid off, without properly accounting for periods when staff were absent due to medical or family reasons.

The lawsuit was filed in the federal court in Oakland, California, following a wave of layoffs announced by Meta in May. The employees claim that while company executives publicly framed the downsizing as a human-driven process, automated systems were operating behind the scenes to harvest data, compare workers, and ultimately determine who would stay and who would go.

According to the statement of claim, Meta scrutinized a broad spectrum of data: Work productivity, code modifications, performance evaluations, usage of artificial intelligence tools, and overall activity on company computers.

Among the allegations, workers were reportedly ranked based on the volume of their interactions with internal AI tools. Those who engaged more frequently with the systems, executed more tasks, and demonstrated higher output allegedly received superior scores.

However, the plaintiffs argue that the systems lacked the capacity to differentiate between an underperforming employee and one who was inactive due to an authorized leave of absence.

Consequently, any employee away on sick leave, maternity leave, parental leave, or family care leave failed to accumulate productivity metrics, utilize internal tools, or generate computer activity. When their metrics were measured against colleagues who remained in the office and continued to work as usual, their scores were severely impacted.

According to the plaintiffs, the practical result was that a metric appearing entirely neutral on paper effectively penalized employees who took leaves of absence that are legally protected.

Approximately half of the plaintiffs were absent from work due to pregnancy, childbirth, parenting, or caring for family members. Others were on leave due to company-approved medical conditions or disabilities.

One employee, according to the lawsuit, returned from a six-month maternity and parental leave – only to receive her termination notice less than two weeks after stepping back into the office.

Another employee claims he was explicitly assured that his medical leave would not negatively impact his performance evaluation. Despite this assurance, he was selected for termination while still on leave. He noted that the only two individuals on his team selected for layoffs were himself and a colleague who was away on maternity leave.

The plaintiffs are not necessarily arguing that the system was intentionally programmed to target women, parents, or employees with medical conditions. Their argument is more nuanced: When a system solely measures activity and output, without understanding the context behind an employee's absence, it can generate discriminatory outcomes even without prior intent.

Meta rejects the allegations. The company stated that the claims are entirely groundless and emphasized that decisions regarding organizational restructuring and workforce management were, and continue to be, made by humans rather than artificial intelligence.

The lawsuit itself does not explicitly claim that an AI system independently finalized the terminations. Rather, the core dispute centers on a different question: To what extent did the automated rankings, scores, and data points influence the managers who ultimate made the final decisions.

The employees are seeking to temporarily halt their terminations until their claims can be thoroughly evaluated in a legal proceeding or arbitration. They contend that the layoffs threaten to strip them not only of their salaries, but also of health insurance, unvested shares, and, in several instances, the work visas required to remain in the country.

The legal action remains in its early stages, and the allegations raised by the plaintiffs have yet to be proven in court.