8,000 Vanish—Who Really Pulled The Trigger?

A yellow folder containing a layoff notice and other documents

A new lawsuit claims Meta let hidden artificial intelligence tools quietly decide who lost their jobs, while the company insists “people, not AI” were in charge.

Story Snapshot

Twenty-six Meta employees say secret artificial intelligence systems helped choose 8,000 workers for May layoffs, allegedly punishing people who took medical or family leave.
The lawsuit says Meta’s internal tools tracked keystrokes, messages, and “token” use, then treated normal gaps for pregnancy or disability as poor performance.
Meta denies the claim and says humans made all workforce decisions, but has not released technical details of the systems.
This may become the first major court test of how far corporations can go using artificial intelligence to manage and fire American workers.

Employees say Meta’s internal AI turned time off into a firing target

Twenty-six current and former Meta workers have filed a federal lawsuit in the Northern District of California, claiming the company used a web of internal artificial intelligence tools to help pick which employees were laid off this May. They say about ten percent of the workforce, roughly 8,000 people, lost their jobs in that round of cuts tied to Meta’s heavy spending on artificial intelligence projects. The plaintiffs are from several states and filed their case anonymously, citing fear of retaliation.

The complaint describes a “constellation” of internal artificial intelligence systems that allegedly scored and ranked employees for termination. According to reporting on the filing, these systems included “Metamate,” a large language model assistant, an internal “second brain” that tracks worker communications and documents, and a productivity score built from keystrokes, screen content, emails, and browser history. The workers argue these tools quietly turned normal life events, like illness or pregnancy, into black marks on their records.

How token counts, productivity scores, and medical leave collided

The lawsuit claims Meta fed performance ratings, calibration scores, productivity and output data, “artificial intelligence–native” ratings, and “artificial intelligence–token consumption” into its systems to help decide who stayed and who was cut. Token consumption is described as a measure of how much an employee uses internal artificial intelligence tools. Workers who were on approved medical or family leave simply could not log the same tokens or activity, so they appeared less “engaged,” even when their leave was protected by law.

The plaintiffs say Meta treated gaps in work history from medical conditions, disability, pregnancy, or caring for family as signs of weak performance, instead of adjusting the metrics. They argue this led the systems to flag many people with disabilities or recent medical leave for layoffs at a higher rate. The lawsuit says this violates federal and state rules that ban discrimination and retaliation against workers who are disabled, pregnant, or use protected leave. It also accuses Meta of failing to test these artificial intelligence tools for bias, despite new rules in places like California and New York City that push companies to do exactly that.

Meta flatly denies AI made the call, but offers few details

Meta has responded with a direct, public denial of the core claim. A company spokesperson said that “workforce management and organisational decisions were and are made by people, not artificial intelligence,” and called the accusations meritless. Chief executive officer Mark Zuckerberg has instead framed the layoffs as a tough financial choice tied to soaring artificial intelligence infrastructure costs, saying more spending on computing power forced hard cuts elsewhere. Meta also says about 7,000 workers were reassigned into new artificial intelligence projects rather than fired in this broader restructuring.

Meta used AI to target workers with medical conditions for layoffs, lawsuit claims https://t.co/aWV8iQp3cw https://t.co/aWV8iQp3cw

— Reuters (@Reuters) July 15, 2026

At the same time, Meta has not released internal documents, code, or detailed logs to show how these decisions were made. The company also has not answered the lawsuit’s specific claims about token counts and productivity scores failing to adjust for medical leave. So far, no named human resources executives or engineers have stepped forward publicly to walk through the process step by step. That means the public is left with two clashing stories and very little hard proof either way, at least until the courts move into discovery.

Why this case matters for American workers and big tech power

Legal experts and reporters say this appears to be the first major case to directly challenge the use of artificial intelligence in firing decisions, not just hiring. It lands at a time when companies across the tech world are blaming artificial intelligence for thousands of job cuts, even as many also report strong profits. One report found nearly forty percent of announced layoffs in May were tied to artificial intelligence, on paper. Some analysts warn that bosses may be “washing” old-fashioned cost-cutting in a shiny tech story to blunt public anger.

For conservative readers, the stakes go beyond one company. If a giant like Meta can quietly run workers’ emails, keystrokes, and health-related leave through opaque algorithms, that raises hard questions about privacy, due process, and basic fairness in the workplace. These systems can become another tool for unaccountable corporate power, with average Americans stuck on the receiving end and no clear way to face their accuser. The coming court fight will test whether existing discrimination and leave laws are strong enough to handle this new kind of digital gatekeeper.