The livestream was supposed to be a standard internal update. Instead, it became a breaking point. During a presentation this week, a Meta employee hijacked the feed with an expletive-laden tirade, demanding that a senior executive be told he was “a piece of sh*t.” As the presenter covered their face in visible distress, the incident laid bare a reality that has been festering inside the company for months: Meta’s massive, three-month-old Applied AI unit is on the verge of collapse.

This isn't just a case of disgruntled staff. It is a structural crisis. Roughly 6,500 engineers and product managers have been effectively conscripted into this unit, tasked with a singular, grinding mission: generating the coding puzzles and technical problems necessary to train Meta’s next generation of AI models. For many, the transition wasn't a choice. It was an ultimatum: join the unit or leave the company.

The 'Draftee' Economy

Employees have taken to calling themselves “draftees.” The process of being moved into the group, according to reports, often arrived via a surprise email, leaving little room for negotiation or pushback. The logic, as articulated by CEO Mark Zuckerberg in leaked internal audio, is that Meta’s own engineers possess “significantly higher” intelligence than the third-party contractors typically used for data labeling.

By pulling from its own internal talent pool, Meta is betting it can solve the data-quality bottleneck that has prevented its models from outperforming humans at complex technical tasks. But the cost of this efficiency is a workforce that feels devalued. “It’s literally the gulag,” one engineer told Wired. “Most people find the work soul-crushing.”

A Culture of Surveillance

The friction extends beyond the nature of the work itself. More than 1,600 employees have signed a petition protesting a program that monitors their clicks and keystrokes to harvest training data. This level of granular surveillance, combined with a management structure that has seen as many as 50 employees reporting to a single manager, has created a pressure cooker environment.

Maher Saba, a 12-year Meta veteran who previously led the Reality Labs division—the unit that famously burned through $83 billion on the metaverse—now heads the Applied AI group. His leadership, reporting up to CTO Andrew Bosworth, has become a lightning rod for the frustration of those who feel they were moved from high-impact product work to glorified data-labeling roles.

What This Means for Meta’s AI Ambitions

Meta’s pivot to AI is not just a strategic shift; it is a total mobilization of the company’s resources. But the internal revolt suggests that the human cost of this mobilization is becoming unsustainable. When your most talented engineers feel they are being treated as disposable data-labeling tools, the risk isn't just low morale—it's the loss of the very talent Zuckerberg claims he wants to retain.

In an internal memo sent Friday, Zuckerberg acknowledged that the recent organizational changes had “caused distress” and admitted to mistakes. He reiterated that “Meta’s north star is to be the best place for the most talented people in the world to make an impact.” Whether that sentiment can bridge the gap between executive vision and the daily reality of the “draftees” remains the central question for the company’s leadership.

Key Takeaways

  • Forced Reassignment: Approximately 6,500 Meta employees were moved into the Applied AI unit with little choice, leading to widespread resentment and the label of "draftees."
  • The Data Bottleneck: Meta is using its own engineers to generate training data because it believes they are more capable than external contractors at solving complex coding tasks.
  • Internal Pushback: A massive petition against keystroke monitoring and a public outburst during a livestream signal that the company’s internal culture is reaching a breaking point.

As the company prepares for its next quarterly update, the focus will shift from the technical capabilities of its models to the stability of its workforce. If the “soul-crushing” nature of the work continues to drive away top-tier talent, the very models Meta is trying to build may never reach their full potential.