Meta’s newest AI organization is three months old and already in open rebellion. According to TechCrunch AI, which is reporting on a new Wired investigation, the company’s roughly 6,500-person Applied AI team has reached a boiling point. The breaking moment: someone hijacked an employee-only livestreamed presentation this week and unloaded an expletive-laden meltdown, demanding that attendees tell a senior Meta AI executive he was “a piece of sh*t.” One presenter reportedly buried their face in their hands.
That outburst wasn’t random. It’s the visible tip of deep anger inside a unit that many of its own members now call a prison.
What actually happened
Here’s the situation TechCrunch AI lays out, drawing on Wired and earlier Business Insider reporting:
- The draft. Thousands of engineers and product managers learned they’d been moved into the Applied AI group through a surprise email. One person described the process on Reddit as “quite random.”
- The job. Their assignment is generating puzzles and coding problems to train Meta’s AI models. As an internal announcement put it, “For agents to understand how people actually complete everyday tasks using computers, we need to train our models on real examples.”
- The mood. “It’s literally the gulag,” one employee told Wired. “Most people find the work soul-crushing,” said another. Many describe a binary choice: join or quit.
The reasoning came straight from the top. In leaked internal audio, CEO Mark Zuckerberg explained why he drafted staff instead of hiring outside contractors. Chief AI officer Alexandr Wang, who sold his data-labeling startup Scale AI to Meta for $14.3 billion, knows the labeling world well. And Zuckerberg added that the average Meta employee has “significantly higher” intelligence than third-party contractors.
Why this matters
What stands out here is the gap between Meta’s ambitions and the people expected to deliver them. The official goal is to close a real weakness: Meta’s models still can’t outperform humans at technical tasks like coding. To fix that, the company needs high-quality human-generated training data. So it conscripted its own engineers to produce it.
This is significant because it shows how the data bottleneck is reshaping the inside of big AI labs. The compute and the talent are only half the equation. The other half is data, and quality human examples are expensive and scarce. Meta’s answer was to repurpose its own workforce as a labeling operation. That tells you how valuable the data has become and how far companies will go to get it.
It also exposes a morale problem that goes beyond one team. According to the report, more than 1,600 Meta employees company-wide have signed a petition protesting a program that monitors their clicks and keystrokes for AI training data. The discontent is broad enough that chief product officer Chris Cox felt he had to address the “brutal” environment on a call with staff this week.
The context
This isn’t Meta’s first expensive bet. The Applied AI team is led by Maher Saba, a 12-year veteran who previously served as a VP in Reality Labs, the division that burned through $83 billion on the metaverse before the company pivoted hard to AI. The new group reports up to CTO Andrew Bosworth. Early on, the structure was so stretched that up to 50 employees reported to a single manager.
Zuckerberg appears to know the situation is bad. In an internal memo Friday, he acknowledged that recent changes had “caused distress” and admitted the company made mistakes it plans to address. He also restated the company line: “Meta’s north star is to be the best place for the most talented people in the world to make an impact.”
What to watch next
The tension between that stated north star and the “gulag” label from inside the building is the story to track. A few things worth watching:
- Whether the promised fixes change the draft-or-quit structure or just soften the messaging.
- Whether other labs follow Meta in turning employees into data generators, or treat this as a cautionary tale.
- Whether the keystroke-monitoring petition forces a policy reversal.
Meta told TechCrunch it had no immediate comment. For the full reporting and leaked details, check the original source.