Meta Sued for Using AI to Target Workers on Leave — The Algorithm That Penalised Pregnancy

Twenty-six Meta employees have sued the company in federal court, alleging that AI tools used to score and rank workers for mass layoffs disproportionately targeted people on maternity leave, disability accommodation, or medical leave — in what may be the first major US case testing whether algorithmic firing violates protected-leave laws.

🔍 THE BOTTOM LINE

Meta’s own AI employee-monitoring program — the one Zuckerberg pitched as training AI “by watching really smart people do things” — became the scoring engine that decided who got sacked. When workers were on legally protected leave, the metrics dried up, the algorithm scored them low, and they were flagged for termination. One plaintiff was notified of her layoff two days before giving birth. The question isn’t just whether Meta did something wrong. It’s whether any company can legally use AI to make firing decisions when the AI can’t tell the difference between “underperforming” and “on maternity leave.”

What the Lawsuit Alleges

The 71-page complaint, filed Monday in the Northern District of California, alleges that Meta used a “constellation of internal artificial intelligence systems” — including AI performance ratings, keystroke monitoring, and activity-tracking data — to select workers for its 8,000-person layoff earlier this year.

“Meta did not assemble the termination list through the considered judgment of managers who knew the work,” the complaint reads. Instead, the plaintiffs allege the company used AI systems “to score, rank and select employees for inclusion on the list.”

The core allegation is structural: Meta’s AI tools gathered data on performance rankings, productivity, and activity metrics. When workers were on medical or family leave, those inputs didn’t exist or were reduced. For people with disabilities, the metrics were necessarily lower. The AI didn’t account for the absence — it treated leave as underperformance.

“The result was that employees who took protected leaves were disproportionately selected for layoff, based on scoring that not only failed to account for their protected leaves, but in effect penalized the employees for exercising their legal rights to these leaves,” the complaint states.

The Workers Behind the Filing

The plaintiffs include a scientist on approved pre-birth pregnancy leave who was notified of her layoff two days before she gave birth. An engineer who said he received a lowered rating because of time off for an injury. A manager on medical leave who was let go 16 days into his time off.

All 26 plaintiffs remain Meta employees until 22 July, when their terminations are set to begin. Their attorneys are seeking a preliminary injunction to stop the layoffs, an independent audit of Meta’s AI tools, and permission for plaintiffs to remain anonymous citing retaliation concerns.

“Once these separations are final, the harms are irreversible: employer-subsidized health coverage lost during pregnancy, postpartum recovery, and active medical treatment; time-bound leave rights extinguished; unvested equity forfeited; and immigration consequences triggered,” the lawyers said in a statement.

The Monitoring Program That Started It All

Meta introduced its AI employee-monitoring program earlier this year, designed to capture keystrokes, mouse activity, browser history, messages, emails, and location data on company devices. Zuckerberg told employees in an internal meeting that the idea was to train AI on their behaviour: “The AI models learn from watching really smart people do things. The average intelligence of the people who are at this company is significantly higher than the average set of people that you can get to do tasks.”

The lawsuit claims the program was quietly launched without employee buy-in, announced through a “low-visibility internal post – made by an engineer rather than a senior leader.” On some teams, employees received no consent prompt and had no way to opt out.

This is the same program that faced internal revolt from more than 1,600 employees who signed a petition saying it violated their privacy, forcing Zuckerberg to pause it in June. The lawsuit now alleges that even during its active period, the data it collected was weaponised against the workers it was monitoring.

Meta disputes the allegations. “These claims lack merit and are not based on facts,” a spokesperson told the Guardian. “Workforce management and organizational decisions were and are made by people, not AI.”

Why This Case Matters Beyond Meta

The US is patchwork-regulating AI in employment. California’s Civil Rights Council secured approval for regulations against AI-related employment discrimination in 2025. Colorado passed SB26-189 addressing automated decision systems. Illinois has HB3773 on AI in hiring.

But none of these laws directly addresses what happens when an AI system that was built for one purpose — monitoring productivity — gets repurposed for another: deciding who gets fired. That gap is what makes this case potentially precedent-setting. If the court finds that Meta’s AI scoring penalised protected leave, it establishes that companies can’t hide behind “the algorithm did it” when the algorithm encodes discrimination.

The case also connects to a broader pattern we’ve tracked: companies rehiring workers after AI-driven layoffs proved costly, and a Chinese court ruling that AI cannot justify worker dismissal. The global trajectory is pointing the same direction — algorithmic firing is facing judicial pushback.

NZ Angle

New Zealand’s employment law has strong protections for parental leave and sick leave under the Parental Leave and Employment Protection Act and the Holidays Act. If a multinational operating in NZ used AI scoring that penalised local employees for taking statutory leave, the Personal Grievance process under the Employment Relations Act would likely apply — and the Employment Relations Authority has historically been hostile to procedural unfairness in dismissals.

The broader question for NZ companies adopting AI workforce tools: are your AI scoring systems accounting for legally protected absences? If not, you may be one lawsuit away from the same problem Meta is now facing.

❓ FAQ

Can companies legally use AI to decide who gets fired? In most US states, there’s no specific prohibition — but existing anti-discrimination laws apply. If an AI system produces discriminatory outcomes (e.g., disproportionately targeting pregnant workers), the company is liable regardless of whether the discrimination was intentional. The Meta case tests whether “the algorithm did it” is a valid defence.

What was Meta’s AI monitoring program? Meta launched a system that tracked employee keystrokes, mouse activity, browser history, messages, and location data on company devices. Zuckerberg said it was to train AI models by observing employee behaviour. Over 1,600 employees signed a petition against it, and Meta paused the program in June 2026.

How does this affect workers in New Zealand? NZ employment law requires fair process in dismissals. If an AI system penalises workers for taking legally protected leave (parental, sick, bereavement), that could constitute unjustified disadvantage or unjustified dismissal under the Employment Relations Act 2000.

What are the plaintiffs seeking? A preliminary injunction to stop the layoffs, an independent audit of Meta’s AI tools, reinstatement, back pay, lost equity, benefits, and damages. They’re also seeking anonymity for plaintiffs and a court order preserving their employment status during arbitration.

Has this happened at other companies? This appears to be the first major US lawsuit specifically alleging that AI scoring penalised protected leave in mass layoffs. But the pattern of AI-driven workforce decisions is widespread — Cloudflare, IBM, and others have faced scrutiny over AI-replacing-worker layoffs.

🔍 THE BOTTOM LINE

Meta built an AI system to watch its employees, then used that same system to decide who to fire. The workers who were on legally protected leave — pregnant, injured, disabled, recovering — had no activity data for the AI to score, so the algorithm marked them as underperforming and slated them for termination. That’s not a bug. It’s the structural consequence of using productivity surveillance as a proxy for human judgment. The lawsuit will determine whether “the AI did it” is a legal shield or a confession.