Judge Refuses to Block Meta Layoffs in AI Discrimination Lawsuit

A federal judge has refused to stop Meta Platforms from laying off 26 workers who claim the company used AI-assisted systems that unfairly penalized employees with disabilities or those who took protected leave.

U.S. District Judge William Orrick denied the workers’ request for a temporary restraining order Friday in Oakland, California. He found they had not shown that the layoffs would cause harm that could not be addressed later if they ultimately prevail.

The ruling does not decide whether Meta’s layoff process violated employment law. Most of the workers’ claims must be handled through individual arbitration under agreements they signed with the company. Those agreements allow employees to ask a court for temporary protection while an arbitrator considers the broader dispute.

A separate request for a preliminary injunctionpreliminary injunctionA preliminary injunction is a temporary court order issued before a case is finally decided that requires a party to take or stop an action while the litigation continues. Federal courts generally consider the likelihood of success, irreparable harm, the balance of equities, and the public interest before granting one.Read more → remains pending. That order could stop the layoffs for a longer period while the dispute continues. Orrick said he could revisit the request after receiving more evidence about whether Meta used AI in the layoff process and how any automated tools affected the selections.

According to the complaint, Meta relied on internal AI-assisted systems, productivity scores and performance data to rank employees during a workforce reduction that affected nearly 8,000 people. The workers claim the process weighed AI token use, output levels and other digital activity, including keystrokes, screen use, emails and browser history.

The lawsuit identifies Metamate, Meta’s internal large language model assistant, and employee-created tools described as a “second brain” that tracked workplace communications and documents. It also points to dashboards that compared workers based on their use of company AI products.

Employees argue that those measures were not paused or adjusted when someone was away from work. As a result, workers on medical, pregnancy, parental or family leave allegedly generated less activity and used fewer AI tools, causing their scores to fall. All 26 workers say they took or requested protected leave, or sought a disability accommodation, during the two years before the layoffs.

The legal dispute rests largely on federal and state laws that protect workers with disabilities and employees who take protected family or medical leave. The Americans with Disabilities Act bars covered employers from treating qualified workers unfavorably because of a disability, including when deciding who will be laid off. The Family and Medical Leave Act gives eligible employees job-protected leave for certain health and family needs.

Those protections do not prevent a company from including an employee with a disability or someone on leave in a broader workforce reduction. An employer may still eliminate the position for a legitimate reason unrelated to the protected condition or absence. The workers claim Meta crossed that line by relying on measures that treated time away from work as lower productivity or weaker performance.

Pregnancy and parental leave are also part of the case. Federal law bars workplace discrimination tied to pregnancy, childbirth and related medical conditions. Employers may also be required to provide reasonable accommodations for pregnancy-related limitations and cannot penalize workers for taking protected leave.

The complaint also raises a form of discrimination known as disparate impact. That claim can apply when a policy appears neutral but places a protected group at a greater disadvantage. A worker does not always have to prove that company leaders intended to discriminate. The focus may instead be on whether the policy caused an unequal result and whether the employer can justify using it as a necessary part of the job or business.

Using software does not remove an employer’s duties under discrimination law. Even when a manager makes the final decision, an employer can still face liability if a ranking system or recommendation helped produce a discriminatory result. California regulations that took effect in October 2025 make clear that state employment protections also apply when companies use AI, algorithms or other automated systems to help make workplace decisions.

Friday’s ruling focused on whether the workers met the standard for immediate court protection, not whether their discrimination claims will ultimately succeed. The employees argued that losing their salaries, stock benefits and employer-supported health insurance would disrupt medical care, pregnancies and family responsibilities. Meta countered that many of those losses could be measured and repaid later if the workers succeed in arbitration.

Orrick concluded that the workers had not shown harm that could not be addressed through money or other relief later. He left open the possibility of revisiting the request after receiving more evidence about Meta’s alleged use of AI in the layoff process.

The preliminary injunction request remains pending, and another hearing is scheduled for August 24. Meta denies wrongdoing and says its workforce and organizational decisions were made by people, not artificial intelligence.