Meta AI-layoff lawsuit puts protected leave under scrutiny

A federal judge has allowed Meta’s planned layoffs to proceed for now while leaving open serious questions about whether AI-assisted workplace metrics may have affected employees who took protected leave or requested disability accommodations.

Judge William Orrick denied an emergency request on July 17 to pause the cuts. The ruling did not decide whether Meta discriminated against workers. Instead, it found that the plaintiffs had raised serious questions about the merits but had not shown the irreparable harm required for a temporary restraining order at that stage.

What the workers allege

The lawsuit was filed July 13 by 26 current and former Meta employees who had been selected for separation and who, within the preceding 24 months, had taken or requested protected leave or sought a disability accommodation.

The plaintiffs allege that Meta used a group of internal AI-assisted and algorithmically assisted systems when selecting workers for a reduction in force. Their filing describes an internal large-language-model assistant, employee-trained tools, productivity scores based on keystroke and device activity, dashboards showing AI-token use, and AI-assisted performance-review and calibration tools.

They contend those systems did not adequately account for periods when employees were away on medical, parental or family leave, or when disability-related accommodations affected their work. The workers argue that protected absences could make employees appear less productive or less engaged when activity and output data were used in rankings or reviews.

Meta announced a reduction in force affecting approximately 8,000 employees, or about 10% of its workforce. Court filings and Reuters reported that many affected workers were scheduled to begin separating from the company on July 22, after being notified of the planned cuts earlier in the year.

Meta disputes the role of AI

Meta says the claims lack merit and are not based on facts. In a declaration submitted to the court, Meta’s human-resources representative said selection decisions were made by human business leaders using documented criteria such as job profile, historical and recent performance, tenure, location, job function, specialized skills and organizational structure.

Meta also told the court that no plaintiff was selected because of leave status, disability or another protected characteristic, and that no selection decision was made by AI. That position directly conflicts with the plaintiffs’ account of how internal tools and workplace data influenced performance evaluations and layoff decisions.

The underlying employment claims are being pursued through private arbitration. The federal court is considering the workers’ requests for temporary injunctive relief while that process moves forward.

What the judge decided — and did not decide

Orrick’s July 17 order was not a merits ruling. He did not find that the allegations were false, and he did not find that Meta violated employment law.

Instead, the judge concluded that the plaintiffs had shown “serious questions” about the merits, a lower showing than proving they were likely to win. But he found that the workers had not established the separate requirement of irreparable harm needed for a temporary restraining order. The layoffs therefore were allowed to proceed for now.

The order says discovery in the arbitration process will be needed to test the competing accounts of how the selections were made. The judge also said the potential immigration consequences for four visa-sponsored plaintiffs could present a different kind of harm, but that he lacked enough information about why those workers were selected to decide the issue at the emergency stage.

Meta was directed to explain by July 23 how and why those four plaintiffs were selected. The plaintiffs may file responsive declarations by August 6.

Why workplace AI is under scrutiny

The case highlights a practical legal question for employers and workers: not simply whether a company used a tool labeled “AI,” but how workplace data was collected, weighted and reviewed before a performance or termination decision.

The Equal Employment Opportunity Commission says existing employment-discrimination laws can apply when employers use automated tools for surveillance, performance assessments, layoffs or termination decisions. Its reduction-in-force guidance advises employers to check whether selection criteria disproportionately affect workers with disabilities or other protected groups and to consider whether the criteria can be adjusted.

That guidance is not a ruling on the Meta dispute and does not establish that Meta’s alleged conduct violated federal law. It does show why employers may face scrutiny over whether protected leave and accommodations were removed or adjusted in performance metrics, whether automated outputs were independently reviewed and what records were kept.

Workers who believe monitoring or automated tools affected a layoff or performance decision may want to preserve layoff notices, performance records, leave documentation and communications. They may also consider contacting the EEOC or seeking advice from an employment lawyer. Employers using these systems should expect questions about selection criteria, disparate effects, documentation and the quality of human review.

What happens next

Meta’s opposition to the plaintiffs’ request for a preliminary injunction is due August 10, and the plaintiffs’ replies are due August 17. Orrick has set a preliminary-injunction hearing for August 24 at 10 a.m. Pacific time.

The hearing will not necessarily resolve the entire case. The central questions remain whether AI-assisted tools or automated metrics materially influenced the selections, how protected absences and accommodations were treated, and whether any use violated federal or state employment law.

Sources

Discover more from Interactive News

Subscribe to get the latest posts sent to your email.