Last week, an email from AI data company Mercor left 5,000 people unemployed overnight.The true cost of the so-called “AI revolution” has finally surfaced.

Overnight Layoffs, AI Firms’ “Pay Cut Restart”

A sudden wave of layoffs caught thousands of AI annotators off guard.The company behind it is Mercor—a firm specializing in data annotation for tech giants like OpenAI, Anthropic, and Meta.Its workers help AI learn to recognize images and understand text, forming the most foundational yet critical layer of AI training.But in mid-November, Mercor abruptly announced the early termination of its “Musen Project” with Meta.This project had employed around 5,000 people. Employees had just heard that clients were satisfied and the project would extend to year-end. Yet, a single email erased everything.Days later, the company notified staff of a new project—Nova.The tasks were nearly identical to Musen, but with reduced wages.The company explained the new project offered more stable tasks and flexible hours. It sounded like an optimization, but was essentially a pay cut.Ironically, Mercor had just secured a new funding round, pushing its CEO Brendan Foody’s valuation to nearly $10 billion.On the surface, it was the rise of a unicorn; behind the scenes, it was restarting the same work at minimal cost.Such practices aren’t new in the industry.Companies like xAI, Scale AI, and Appen have all gone through similar cycles: projects end, hordes of contract workers are laid off, then “rehired” under new names and lower pay.Mercor simply brought the industry’s unspoken rules into plain sight.It happened swiftly and quietly. While AI firms continued their upward trajectory in capital markets, the very people propelling their growth were silently replaced behind the screens.

Workers on the AI Assembly Line: “Want to Resist, But Can’t Afford To”

When Mercor sent out those termination notices, many employees were stunned.One annotator shared in an interview:”They told us in the group chat that clients were satisfied and the project would be renewed. I breathed a sigh of relief, only to be told the project was ending the next moment.”Someone wrote on Reddit about their experience: Waking up at 3 a.m. to a termination email, logging into the work platform the next day to find their task page wiped clean, leaving only a message saying, “Thank you for your efforts.”Within days, a new offer arrived.The company called it a “project transition,” inviting them to join the new “Nova” project. The only catch? Hourly pay dropped from 21to16.Mercor explained in the email that the adjustment aimed to “ensure task stability” and “improve collaboration efficiency.”In other words: doing the same work for less money.An employee who accepted the new contract said:”We want to resist, but we can’t afford to. We need that stable income, even if it feels demeaning.”Most signed the new contracts. Some had children to support, others needed the money for student loans, and some couldn’t find alternative jobs.They weren’t Mercor’s full-time employees—just temporary contractors. No insurance, no paid leave, no bargaining power.They knew they were being exploited, but had no choice.

CEO’s “Human Prosperity,” Not Meant for Them

If you only look at capital markets, AI seems to be in a golden age.U.S. Commerce Department data shows that in the first half of 2025, the tech sector contributed 92% of the nation’s GDP growth.Yet, during the same period, tech layoffs hit their highest level since 2003.Amazon just cut 14,000 corporate roles, while Google and Meta continue “team optimizations.”It’s paradoxical—profits are up, jobs are down; companies are expanding, yet people are being erased.Yet CEOs remain eager to paint a future where “AI makes humanity happier.”OpenAI’s Sam Altman once said in a speech:”We are moving toward an era where human prosperity will reach unimaginable heights.”But in the worlds of Mercor, Appen, and Scale AI, AI hasn’t made people freer—it has made them cheaper.Data annotators are the construction workers of the AI world. Their labor creates value for models, yet they’re excluded from the prosperity.We can’t help but ask: Who exactly does Altman’s “humanity” refer to?Investors? Developers? Or those workers who received termination emails at 3 a.m., struggling to sign new contracts?If human prosperity means a tiny minority controlling computing power and capital, while the majority are re-priced by algorithms, then this utopia may already be teetering on the edge of hell.

Called “Freelancers,” But Actually Abandoned

One term frequently used in the AI industry is “flexible employment.”Annotators are no longer employees—they’re “independent collaborators” or “freelancers.” It sounds easier, freer, like a new identity for the tech era.But in reality, this “freedom” is more like abandonment.Companies like Mercor, Scale AI, and Appen break work into fragmented tasks, outsourcing them to global contractors, while shedding all responsibility.No insurance, no holidays, no base salary—not even formal layoffs. Just an email, a deactivated login.This structure traps people in a paradox: the more they seek stability, the more they rely on unstable work.