5,000 Employees Cut Overnight! Mercor’s Salary-Reduction Rehire Strategy Sparks Outrage

The world’s excitement for AI has never slowed. Tech giants are burning money at unprecedented levels, analysts claim AI is responsible for nearly 90% of U.S. GDP growth, and media headlines boast about “the dawn of a new era of human prosperity.”

Yet beyond this celebration, there is a quieter story—one of people disappearing from the landscape, without applause or acknowledgment.

Last week, a single email from AI data company Mercor left 5,000 workers jobless overnight. The wave of the so-called “AI revolution” finally revealed its price.

Fired overnight — and brought back for less

The sudden layoff struck thousands of AI data labelers with no warning.

Mercor, a company specializing in annotation work for OpenAI, Anthropic, and Meta, relies on these workers to teach AI to recognize images and understand language—work that is foundational, essential, and invisible.

But in mid-November, Mercor abruptly announced that the “Musen Project,” done in partnership with Meta, had ended ahead of schedule. Ironically, just days earlier, employees had been reassured that customer satisfaction was high and the project would continue through the end of the year.

Then came the email—and their work simply ceased.

Days later, Mercor announced a new project—internally named “Nova.” The tasks were nearly identical, but the pay was lower. The company described it as a move toward “greater stability and flexibility.” In reality, it was a pay cut disguised as optimization.

Adding to the irony, Mercor recently secured a new round of funding and now boasts a valuation of nearly $10 billion. Publicly they are celebrated as a rising unicorn—privately they are rebooting the same work at a lower cost.

This pattern is nothing new. Companies like xAI, Scale AI, and Appen have repeatedly laid off armies of contractors at the end of one project, only to rehire them under new terms, new rates, and new branding.

The difference this time is that Mercor has dragged the industry’s quiet exploitation out into the open. It all unfolded quickly, quietly, and clinically. While AI companies climb upward in valuation, the people behind the curtain—the ones labeling the data—are quietly replaced.

On the AI assembly line: “We want to protest, but we need the money”

When the contract termination emails landed, many workers simply froze in disbelief.

One annotator shared:
“They told us the client was happy and the project would continue. I finally relaxed… and then the next thing I know, I’m out.”

Another wrote on Reddit: they received the termination email at 3 a.m., and by the next morning their task dashboard was wiped clean—leaving only a single sentence: “Thank you for your contribution.”

Then came the invitation to join “Nova.” Same job. Lower pay. Hourly rates dropped from $21 to $16.

Mercor justified the move by saying the adjustment would “ensure task consistency” and “improve operational efficiency.” In other words: same work, less money.

One worker accepted, reluctantly:
“We want to resist—but we need the income. Even if it feels degrading.”

So they signed. Most of them did. Some have children to feed. Some have student debt. Some simply cannot find alternative work.

They are not full-time employees. They are contract labor—no health benefits, no paid leave, no room to negotiate. They know it’s exploitation. And they accept it because they must.

The CEOs talk about “human prosperity”—but not for everyone

On paper, AI is riding a golden age. According to the U.S. Department of Commerce, the tech sector contributed 92% of all GDP growth in early 2025.

Yet simultaneously, tech layoffs have hit record-high levels.

Amazon just cut 14,000 corporate roles. Google and Meta continue their “workforce optimizations.” There is a glaring contradiction: profits rise, and jobs evaporate.

Still, CEOs continue proclaiming a vision of a flourishing human future.

As Sam Altman of OpenAI declared:
“We are moving toward a future of human prosperity beyond imagination.”

But inside the ecosystems of Mercor, Scale AI, and Appen, AI is not making life freer—it’s making labor cheaper. Labelers are the construction workers of the AI world. Their work creates value—but they are excluded from the prosperity.

So we have to ask: when Altman says “humanity,” who exactly is included?

Investors? Founders? Model engineers? Or the workers reading termination emails at 3 a.m., forced to sign a lower-pay contract by sunrise?

If human prosperity means a tiny elite hold the compute and capital—while the many are algorithmically devalued—then this utopia is already far too close to dystopia.

“Freelancers” in name—abandoned in practice

The AI industry loves the term “flexible work.” Data annotators are no longer employees—they are “independent collaborators,” “freelancers,” “global partners.” It sounds empowering, modern, and free.

In reality, this “freedom” is abandonment.

Mercor, Appen, and Scale AI reduce work into micro-tasks distributed to a global workforce—while corporations shed responsibility. No insurance, no leave, no base salary. And termination is not even termination—just a login that no longer works.

This creates a paradox: the more workers crave stability, the more they are locked into unstable work.

AI companies wrap uncertainty in the language of flexibility. They replace employment with “collaboration.” They replace management with algorithms.

Behind the screen, annotators continue training the machines, making AI smarter.

Yet they themselves have been erased from the system’s memory.

The AI world is moving forward—yet their wages have slipped backward, back to levels from a decade ago.