Meta is facing renewed legal scrutiny after four employees holding company-sponsored work visas argued that layoffs could force them to leave the United States if they fail to secure a new employer within the limited grace period. Although a federal judge declined to temporarily block Meta’s mass layoff plan, he acknowledged that the immigration-related risks facing these workers warranted closer examination.

The lawsuit is part of broader litigation initiated by 26 current and former Meta employees, who allege that the company used artificial intelligence systems to help determine who would be laid off. Many of the plaintiffs allege that they were on maternity or sick leave, or were utilising workplace accommodations for disabilities, at the time of their selection.

In a ruling issued on July 17, District Judge William Orrick denied the workers’ request for a temporary restraining order that would have halted the layoffs. However, he made an exception for the four plaintiffs whose work visas are sponsored by Meta.

The judge noted that the immigration consequences for these workers are “more than mere speculation.” “According to court documents, once their employment ends, they could have a grace period of just 60 days to find another employer willing to sponsor their visa, or else risk losing their legal status in the United States. Consequently, Judge Orrick ordered Meta to explain, by July 23,” how and why the four plaintiffs identified as ‘Doe’—holding Meta-sponsored work visas—were selected for layoff. In court documents, the four employees are identified as Does 4, 9, 15, and 26. The court noted that while economic losses—such as salary, stock, or health insurance—can be compensated later through arbitration, the immigration consequences for visa holders constitute a distinct type of harm that is not easily reversed.

According to a famous publication the class-action lawsuit filed in the Northern District of California claims that Meta used a network of internal artificial intelligence tools to rank employees during this year’s workforce reduction, which affected approximately 8,000 workers.

The lawsuit states that the company used AI-generated performance scores, along with work activity data—including productivity metrics and other internal indicators—to determine which employees would be included in the layoffs. The plaintiffs argue that this process unfairly penalised workers on protected leave, as they naturally generated less work activity during that period.

One employee recount being notified of her layoff just two days before giving birth, while on authorised maternity leave. Another engineer claims his performance rating dropped after taking time off to recover from an injury, while a manager asserts he was laid off after being on medical leave for just over two weeks.

The lawsuit contends that, rather than relying primarily on managers familiar with the employees’ work, Meta used AI systems to “score, rank, and select employees” for layoffs.

The employees are seeking reinstatement, back pay, the restoration of lost benefits and stock, and an independent audit of Meta’s AI-based decision-making process. Meta has rejected the accusations. A company spokesperson recently told a famous publication that the claims are “baseless and not grounded in fact,” adding that organisational and personnel decisions “were and are made by people, not AI.”