Meta AI Confirms Terminated Scientist 'Was Not Generating Tokens'; Notes Model Has No Field for Why

MENLO PARK— Meta's AI-powered performance-ranking system, which twenty-six current and former employees allege illegally selected workers on medical, pregnancy and parental leave for termination during May's 8,000-person layoff, achieved a precision rate the company called "industry-leading" in its Monday response to the federal suit filed in Oakland.

"The tool correctly identified which employees were not using our AI productivity suite," a spokesperson said. "That is the metric."

Court documents show the model scored workers daily on a composite index that weighted calendar density, message-response latency and AI-token consumption across Meta's internal tooling. One plaintiff, a research scientist, held a score of 74 before her approved leave began. It reached 12 by the end of her second week away from a computer. Her termination notice was issued six hours before her delivery.

"She was not generating tokens," the spokesperson said. "The model does not have a field for why."

Meta said it has since refined the system. Workers on documented leave will now receive a protective baseline score of 15, which the company described as "a meaningful floor." Fourteen employees whose scores rested exactly at 15 received reduction-in-force notices the following Monday.

Asked whether the company planned to add a field for why, Meta said it was examining all of its review processes.

The examination, a spokesperson confirmed, would be conducted by the model.