The algorithm treated medical leave as a data point signaling expendability.
In Oakland, twenty-six former Meta employees have brought a federal lawsuit alleging that the company deployed artificial intelligence to identify and remove workers with disabilities or medical conditions during its sweeping 2025 layoffs. The case asks a question that grows more urgent as algorithms assume greater authority over human livelihoods: when a machine shapes the list and a person signs it, where does responsibility truly reside? At its heart, this is not merely a dispute about one company's workforce decisions, but a reckoning with whether civil rights law — written for a world of human prejudice — can reach the encoded biases of the systems we have built to replace human judgment.
- Twenty-six workers, many managing chronic illness, surgery recovery, or pregnancy-related conditions, allege they were quietly sorted out of their careers by an algorithm that read their medical absences as signs of expendability.
- Meta's AI system allegedly weighted productivity metrics and AI token usage in ways that structurally penalized anyone who had stepped away from work for protected medical reasons — turning absence into a data-point for termination.
- Meta has pushed back firmly, insisting that humans, not machines, made all workforce decisions — a defense that frames the AI as merely advisory, but which the plaintiffs argue misunderstands how algorithmic filtering shapes every choice that follows.
- The case now hinges on discovery: plaintiffs' lawyers will seek the system's training data, its weighting logic, and any internal awareness that the algorithm would disproportionately harm disabled workers.
- The outcome carries stakes far beyond Meta — a plaintiff victory could mandate algorithmic bias audits across the tech industry, while a Meta win might entrench the 'human clicked last' defense as a shield against discrimination liability.
Late Monday, a federal courthouse in Oakland received a filing with a particular sting: twenty-six former Meta employees, filing anonymously from six states and Washington, D.C., allege that the company built an AI system designed — whether by intent or by logic — to find and fire people with disabilities.
The alleged mechanics are both simple and damning. When Meta announced it would cut roughly 8,000 jobs earlier this year, the lawsuit claims its AI did not evaluate workers on merit alone. Instead, it weighted productivity metrics and AI token usage — measures that would naturally disadvantage anyone who had stepped away for surgery, chemotherapy, or the management of a chronic condition. Medical leave, the plaintiffs argue, was not treated as a protected status but as a data signal marking someone as expendable.
The legal foundation draws on decades of civil rights protections: federal and state laws prohibiting discrimination on the basis of disability, retaliation for medical leave, and penalties for pregnancy-related absences. What distinguishes this case is the mechanism — not a manager's conscious prejudice, but an algorithm that may have encoded bias into its very architecture.
Meta responded swiftly, with a spokesperson stating that the allegations lack merit and that 'workforce management and organizational decisions were and are made by people, not AI.' It is a familiar defense in the age of algorithmic accountability — the company is arguing, in effect, that the machine was advisory and humans held final authority.
But the lawsuit presses a harder question: what does human decision-making mean when the information presented to humans has already been filtered and ranked by an algorithm? If the system flagged workers on medical leave as lower-priority, and those workers were disproportionately terminated, does it matter who clicked the final button?
The case will likely turn on discovery — the examination of how the AI was trained, what it weighted, and whether anyone inside Meta understood it would disadvantage disabled workers. If the plaintiffs prevail, it could compel tech companies to audit algorithmic bias in hiring and firing as standard practice. If Meta wins, it may entrench the principle that human sign-off is sufficient to insulate a company from discrimination claims, no matter what the algorithm did first. For twenty-six workers, this is a fight for accountability. For everyone else, it is a test of whether the law can keep pace with the tools now deciding who works and who does not.
Late Monday in Oakland, a federal courthouse received a filing that names Meta as a defendant in a discrimination case with a particular sting: the company, the lawsuit alleges, built a machine to find and fire people with disabilities. Twenty-six former employees, filing anonymously from six states and Washington, D.C., are accusing the tech giant of deploying AI software that systematically identified workers on medical leave or living with disabilities, then marked them for termination during the company's sweeping job cuts earlier this year.
The mechanics of the alleged scheme are straightforward and damning. Meta announced in May that it would eliminate roughly 10 percent of its global workforce—nearly 8,000 people—with additional cuts to follow. According to the lawsuit, the company's AI system did not evaluate workers on merit alone. Instead, it weighted factors like productivity metrics and something called AI token usage, measures that would naturally disadvantage anyone who had stepped away from work for medical reasons. A person recovering from surgery, undergoing chemotherapy, or managing a chronic illness would show lower productivity numbers simply by virtue of their absence. The algorithm, the plaintiffs argue, treated medical leave not as a protected status but as a data point signaling expendability.
The legal theory underlying the case is rooted in decades of civil rights law. Federal statutes and parallel state laws prohibit employers from discriminating against workers because of disability, from retaliating against those who take medical leave, or from penalizing pregnancy-related absences. The plaintiffs contend that Meta violated all three categories of protection. What makes this case distinctive is the mechanism: not a manager's conscious bias, but an algorithm that encoded bias into its very logic.
Meta's response has been swift and categorical. A company spokesperson said on Tuesday that the allegations lack merit, and more pointedly, that "workforce management and organizational decisions were and are made by people, not AI." The statement attempts to draw a bright line between the tool and the decision-maker, suggesting that humans, not machines, bear responsibility for who stays and who goes. It is a familiar defense in the age of algorithmic accountability—the company is saying, in effect, that the AI was merely advisory, that final choices remained in human hands.
But the lawsuit raises a question that has begun to haunt tech companies and their legal departments: what does human decision-making mean when the information fed to humans has already been filtered, ranked, and shaped by an algorithm? If the AI system flagged workers with medical leave as lower-priority candidates, and those workers were disproportionately terminated, does it matter whether a person clicked the final button? The plaintiffs seem to be arguing that the algorithm did the real work of selection, and that Meta's reliance on it constitutes discrimination regardless of who formally approved the terminations.
The timing of the lawsuit is significant. Meta's layoffs were not unusual in the tech industry—many companies cut headcount in 2024 and 2025—but the scale was notable. Eight thousand jobs represents a substantial portion of the company's workforce. If the allegations are true, the use of AI to identify targets would represent a systematic approach to discrimination, not isolated incidents of bias. It would suggest that the company built efficiency into its cruelty, that it automated the process of deciding who deserved to keep their livelihood.
The case will likely turn on discovery—the process by which both sides examine the company's internal documents, code, and decision-making records. Lawyers for the plaintiffs will want to see how the AI system was trained, what data it used, how it weighted different factors, and whether Meta's engineers or executives knew that the system would disadvantage disabled workers. Meta will argue that the layoffs were driven by legitimate business needs and that any disparate impact on disabled workers was coincidental, not intentional.
What happens next will matter beyond Meta. If the plaintiffs prevail, or if the case settles with significant concessions, it could force tech companies to rethink how they use AI in workforce decisions. Audits of algorithmic bias in hiring and firing could become standard practice. Conversely, if Meta wins, it may signal that companies can shield themselves from discrimination liability by claiming that humans, not algorithms, made the final call. The case sits at the intersection of disability rights, employment law, and the growing power of artificial intelligence to shape human lives. For the twenty-six workers who filed it, the lawsuit is a chance to hold a powerful company accountable. For the broader economy, it is a test of whether the law can keep pace with the tools that now decide who works and who does not.
Citações Notáveis
Workforce management and organizational decisions were and are made by people, not AI.— Meta spokesperson
A Conversa do Hearth Outra perspectiva sobre a história
Why does it matter whether an AI system recommended the layoffs versus a human making the choice directly?
Because if the algorithm did the real filtering—if it identified disabled workers as lower-priority based on productivity metrics skewed by medical leave—then the human decision becomes almost ceremonial. The damage is already baked in.
But Meta says humans made the decisions. Doesn't that count for something?
It depends on what information those humans had. If the AI presented them with a ranked list of candidates to cut, and that list was already biased, then human approval doesn't erase the bias. It just launders it.
What would the plaintiffs need to prove to win?
They'd need to show that the AI system was designed or operated in a way that predictably disadvantaged disabled workers, and that Meta knew or should have known this would happen. Internal documents about how the system was built would be crucial.
Could this change how other tech companies do layoffs?
If the plaintiffs win, yes. Companies would have to audit their algorithms for bias before using them in workforce decisions. If Meta wins, it sends a message that you can use AI as long as you have a human sign off at the end.
What's the human cost here?
Twenty-six people lost their jobs, possibly because a machine flagged them as expendable based on time they needed to recover or manage illness. That's not just a legal question—it's about dignity and whether disability makes you disposable in the eyes of your employer.