Twenty-six Meta employees filed a lawsuit in a U.S. federal court accusing the company of using artificial intelligence (AI) systems to analyze work data and rank employees for layoffs, resulting in unfair impacts on disabled workers, employees on sick leave, maternity leave, and family care leave. Meta denies these allegations, affirming that all decisions were made by humans, not AI.
A group of employees from Meta Platforms filed a lawsuit in a U.S. federal court accusing the parent company of Facebook, Instagram, and WhatsApp of secretly using AI systems to evaluate and select employees for a large wave of layoffs. The system is alleged to be biased and improperly targeted disabled employees, those on maternity leave, and those on sick leave. This marks a landmark case challenging the use of AI to dismiss workers in a major tech company.
The 71-page complaint filed in a federal court in Oakland, California, states that 26 anonymous plaintiffs, fearing retaliation, jointly accuse Meta of relying not on the discretion of managers or supervisors familiar with the work, but on an "internal AI network within Meta" to score, rank, and select employees for termination.
The AI system includes a large language model assistant called "Metamate" and a "Second Brain" system trained to monitor communications, documents, and productivity scores by surveilling employee behavior through company devices, such as tracking keyboard typing frequency, mouse movements, screen content, emails, and web browsing history.
This system was quietly launched early in the year to have AI learn from the work habits of the company's most productive employees, following CEO Mark Zuckerberg's vision. However, it caused strong employee backlash, with over 1,600 workers signing a privacy protest petition, leading Zuckerberg to temporarily suspend the surveillance project in June.
The plaintiffs state that because the AI continuously calculates scores based on performance and token usage, employees who legally took sick leave, family care leave, or requested workplace accommodations due to disabilities had their scores drop or disappear during their absence. This led the AI to interpret these employees as underperforming, automatically penalizing those exercising their legal rights.
The complaint cites notable cases, such as a female scientist on approved maternity leave who received a layoff notice just two days before her due date; an engineer whose evaluation score plummeted because he took time off for injury recovery; and a manager on sick leave who was terminated after only 16 days off.
These layoffs are part of a major reduction of 10% of Meta's workforce, around 8,000 positions globally, implemented earlier this year to restructure the company towards full AI investment. The affected employees were informed that their positions would be eliminated starting 22 July.
Meta's spokesperson promptly responded to the allegations, stating, "These claims are baseless and not grounded in fact. Personnel management and organizational decisions are all made by humans, not AI."
The plaintiffs, spread across six states including California, New York, and the District of Columbia, have requested the court to issue a temporary restraining order to halt the planned layoffs on 22 July while the case proceeds to private arbitration. They also seek appointment of an independent auditor to review Meta's AI systems.
Plaintiffs' attorneys emphasize that if the layoffs proceed, the harm to employees will be "irreparable," including loss of company-subsidized health insurance during pregnancy, postpartum recovery, ongoing medical treatment, forfeiture of unvested company stock, and impacts on the immigration status of foreign employees.