Twenty-six former Meta employees are suing the company for allegedly using a biased AI tool that “unfairly selected” people for termination after taking sick leave. Reuters I will report it. All plaintiffs in the case were affected by Meta’s latest round of layoffs, which cut approximately 8,000 employees to offset investments in AI and data center infrastructure.
Meta is accused of using a “series of artificial intelligence systems” to select which employees to cut. These allegedly included an in-house AI assistant called ‘Metamate’, a ‘second brain’ agent trained by employees, an AI token usage dashboard, and ‘keystroke and activity monitoring data’. The alleged goal was to identify and rank employees based not only on their performance and productivity, but also on how “AI native” they were, the number of AI tokens they used, and more. The problem, the lawsuit alleges, is that Meta’s system did not account for employees on family or medical leave, or whose disabilities could impact production, even though they were unable to use more tokens or utilize other AI tools.
Engadget reached out to Mehta for comment on the lawsuit and the use of AI. In a statement given to Reutersthe company said the lawsuit lacks merit because terminations always involve human intervention. “Employee management and organizational decision-making has been and is still being done by humans, not AI,” Mehta said.
A key claim in the lawsuit is that each of the plaintiffs “obtained, requested, or was approved to take legally protected leave, was prevented from taking protected leave, or requested or obtained a reasonable disability accommodation” within 24 months of their termination. Laws such as the Family Leave Act and the Medical Leave Act explicitly prohibit companies from considering employees’ access to protected leave as part of hiring decisions. At the state level, the California Family Rights Act also prohibits this, and the state’s Fair Employment and Housing Act “prohibits the use of automated decision-making systems that create differential impact discrimination on the basis of gender, including disability or pregnancy.”
The former employees are asking the court to prevent Meta from completing the job cuts “until an independent audit of the algorithm-assisted selection process is conducted.” Under the terms of Meta’s employment agreement, the plaintiffs intend to pursue their claims in arbitration.
Reuters Meta first reported in April 2026 that it was recording staff keystrokes, mouse movements, and clicks to train its AI. In addition to sounding invasive and creepy, the program reportedly violates European Union privacy laws. Meta chose to suspend its AI training program after it was discovered that employees had access to private chats and transcripts recorded by the system. Given the humiliation of the company’s original plans, the idea that similar surveillance data could have been used to cut jobs is not particularly surprising.
