Meta sued by 26 workers over AI-assisted layoff discrimination

On Monday, 26 unnamed current and former Meta employees, represented by a coalition of attorneys, filed a lawsuit in the U.S. Northern District Court of California accusing Meta of discrimination in its May layoff round. The round cut 10% of Meta's workforce, and the plaintiffs were among those terminated. The lawsuit alleges that Meta violated protected-leave laws and discrimination statutes related to pregnancy, disability, and other protected statuses.

According to the legal complaint, Meta's "constellation of internal artificial-intelligence systems" failed to account for approved absences when determining which employees to cut. The systems evaluated workers on performance ratings, calibration scores, productivity and output metrics, what the company terms "AI-native" ratings, and AI-token consumption. Attorneys argued that these metrics—by design—cannot be accumulated by employees on protected medical or family leave, or whose output is reduced by a disability. Token consumption, in particular, has become a proxy for general AI usage at the company, and the plaintiffs allege Meta weaponized this and similar metrics to target certain workers.

Meta denied the allegations through a spokesperson, telling CNBC in an email that the "claims lack merit and are not based on facts" and that "workforce management and organizational decisions were and are made by people, not AI." The plaintiffs are seeking a preliminary injunction to maintain their employment status at Meta pending an independent audit of the algorithmically assisted selection process and resolution of the claims in arbitration.

The lawsuit arrives in a context of heightened legal scrutiny of AI in employment. Roughly a month earlier, a federal judge in California ruled that tech firm Workday must face discrimination claims regarding its AI-powered job screening services, which allegedly violated state and federal discrimination laws. Workday denied the allegations and said its AI recruiting software examines only job qualifications, not protected traits like race, age, or disability, and that it rigorously tests products through its Responsible AI program.

The lawsuit reflects a broader pattern of legal challenges to AI-assisted employment decisions. The 26 plaintiffs allege that Meta's constellation of AI systems created a structural disadvantage for workers with disabilities or on protected leave—not through explicit rules, but through metrics that inherently cannot accrue value for such employees. Token consumption, mentioned in the complaint, has become a shorthand measure of AI engagement at Meta and other tech companies; the plaintiffs argue this metric, combined with others, functionally screened out protected workers.

Meta's defense—that humans, not AI, made the final decisions—mirrors arguments made by other tech companies facing similar litigation. A federal judge in California rejected a comparable argument from Workday roughly a month before this Meta lawsuit was filed, ruling that Workday must face claims about its AI-powered job screening. That precedent suggests courts are willing to hold companies accountable for AI systems' effects even when humans retain formal sign-off authority. The lawsuit also asks for a preliminary injunction and an independent audit, remedies that could impose significant operational constraints if granted.