Meta says managers decided who got laid off in May. Twenty-six former employees allege it was the AI.
In a lawsuit filed Monday, the group alleges that Meta used a collection of internal AI tools to score, rank, and build the termination list for its May layoffs, which cut roughly 8,000 workers. The suit names an internal AI assistant called Metamate, employee-trained AI agents, and dashboards tracking individual AI token usage as the systems that fed the rankings.
The core claim is that the system never excluded employees on medical leave. Their time away was treated as lower performance, and they were "disproportionately selected for layoff" as a result.
A Meta spokesperson denied the claims, saying "workforce management and organizational decisions were and are made by people, not AI."
That defense has already been challenged in a federal courtroom. In Mobley v. Workday, a job applicant sued Workday's AI hiring platform for discriminating against him on the basis of race, age, and disability. Workday made the same argument, that it was just a tool, not an employer making hiring decisions. Judge Rita F. Lin ruled that the software was not "simply implementing in a rote way the criteria that employers set forth" but was "participating in the decision-making process." She allowed discrimination claims against the AI vendor to proceed.
Mobley was about hiring. The Meta lawsuit is about firing. Between the two, there is now no stage in the employment lifecycle where AI-driven decisions are safe from legal challenge.
The bias concerns aren't hypothetical. A Stanford study of 3.37 million real job applications found that even AI screening systems designed to be fair produced significant racial disparities when examined position by position. About 4% of people who applied to 10 algorithm-screened jobs were rejected from all of them at rates higher than chance could explain. The researchers called it "algorithmic blackballing." When an AI trains on your existing workforce to learn what good performance looks like, it repeats whatever patterns are already there. That applies to who gets fired just as much as who gets hired.
Only 26 of the roughly 8,000 laid-off workers are suing so far. But if the court finds that AI tools were used to build the full layoff list, the legal exposure covers everyone who was cut.
The "humans decided, not AI" defense is the one every company will reach for when AI-driven workforce decisions get challenged. A federal judge has already refused to dismiss a case challenging it. If you work at a company of any real size, there is a reasonable chance some system has already evaluated you in ways you've never seen and you had no opportunity to challenge. Unlike your credit score, no law currently requires your employer to show you that number before it's used against you. These two cases are the start of that changing.
