Meta’s AI overhaul has turned its Applied AI unit into a “soul-crushing gulag,” according to internal reports, while its CTO admits morale is at a 20-year low after waves of layoffs and forced reassignments. The company’s AI push—backed by $83 billion in Reality Labs spending—is now colliding with a workforce in revolt, with employees calling the work “mechanical” and “not creative.”
Meta’s latest round of layoffs in May—affecting 8,000 employees, or 10% of its workforce—was explicitly tied to funding its AI ambitions. But the fallout has been swift: a leaked internal meeting saw an employee livestream a tirade against leadership, calling a senior AI executive “a piece of sh*t.” Meanwhile, a petition signed by over 1,600 workers protests a program monitoring keystrokes and mouse movements for AI training data. The chaos has even prompted Meta’s chief product officer to call the environment “brutal,” comparing it to “running a marathon in the middle of a hailstorm.”
How Meta’s AI Unit Became a “Gulag”
Meta’s Applied AI unit—formed in March to support its AI research—has become a flashpoint for employee discontent. The unit, now comprising roughly 6,500 engineers and product managers, was assembled through a process employees describe as “drafting.” According to TechCrunch, workers were reassigned without choice, tasked with generating coding puzzles and technical problems to train AI models. One employee told the New York Post the work was “mechanical and not creative,” adding, “You have zero purpose in life all of a sudden.”
Employees describe the unit’s structure as chaotic, with up to 50 workers reporting to a single manager—a setup that has exacerbated frustration. The unit’s leader, Maher Saba, a 12-year Meta veteran, oversees the team under Meta CTO Andrew “Boz” Bosworth. The work itself is seen as soul-crushing: instead of building products, employees are fed into a pipeline of repetitive, low-value tasks. As one worker put it to Wired, “It’s literally the gulag.”
The resentment runs deeper than the AI unit. A broader petition against Meta’s keystroke-tracking program—used to gather data for AI training—has garnered over 1,600 signatures, signaling widespread unease. The program, which monitors employee activity to improve AI models, has been met with resistance, with workers calling it an invasion of privacy. Meta’s chief product officer, Chris Cox, addressed the issue in a recent call, acknowledging the “difficult” and “brutal” conditions. He reportedly told employees, “It is neither god, nor is it the devil,” comparing the company’s AI push to “running a marathon in the middle of a hailstorm.”
Morale Crashes as Bosworth Admits “Worst in 20 Years”
Meta’s CTO, Andrew Bosworth, has publicly acknowledged the severity of the morale crisis in a rare admission of failure. During an internal “Tuesdays with Boz” session on June 2, Bosworth described morale as “maybe not the worst it’s ever been in 20 years here, but it’s probably up there. It’s definitely up there,” according to Business Insider. He compared the current climate to the fallout from the 2016 Cambridge Analytica scandal, a low point for the company.
Bosworth’s memo to employees on Monday admitted Meta had “done an atrocious job explaining the vision” behind its AI restructuring and had “undermined the trust” of its workforce. He pledged greater transparency, support for career growth, and a return to the company’s original culture. Yet the damage is already done: employees report feeling “drafted” into AI roles with no real choice, and many describe the work as soul-crushing. The memo also announced increased budgets for travel, events, and office perks—small gestures in the face of widespread discontent.
For more on this story, see Meta cuts 8,000 jobs in global layoffs, offers 4 months’ severance.
Bosworth’s comments come after Meta’s latest layoffs in May, which followed a “Year of Efficiency” in 2023 that saw another 10,000 roles eliminated. The company has now laid off roughly 20% of its workforce since late 2022, all to fund its AI push. Yet despite the cuts, Meta’s AI efforts remain underwhelming compared to competitors. As Futurism reported, Mark Zuckerberg’s attempt to lift spirits with a companywide AI hackathon in July was met with derision. One employee wrote in an internal message, “I’m literally preoccupied with keeping the lights on for my team. I have no incentive to participate.”
Why the AI Push Is Backfiring
The root of Meta’s problems lies in its AI strategy: a $14.3 billion acquisition of Scale AI, a data-labeling startup, and a $83 billion burn rate in Reality Labs. The company’s shift toward AI has required massive workforce reductions to fund infrastructure, but the execution has been chaotic. Employees reassigned to AI teams report feeling like “draftees,” with little autonomy or purpose. The work—generating coding puzzles and labeling data—is seen as menial, a far cry from the creative, high-impact roles many joined Meta for.
Meta’s AI ambitions are also out of step with its workforce. The company’s Applied AI unit was assembled to address a critical gap: its AI models still lack the technical knowledge to outperform humans in tasks like coding. But the solution—reassigning employees to data-labeling roles—has backfired. As Yahoo Finance reported, Meta’s CTO acknowledged the company had “undermined the trust” of its employees by failing to explain the vision behind the AI push. The lack of clarity, combined with forced reassignments, has left many feeling betrayed.
The contrast with competitors is stark. While Meta struggles with internal revolt, companies like Google and Microsoft have managed to integrate AI into their workflows without triggering similar backlash. Meta’s approach—centralizing AI efforts under a single, poorly structured unit—has created a sense of isolation among employees. The Applied AI team’s structure, with up to 50 workers reporting to one manager, has only worsened the sense of disconnection.
What Happens Next?
Meta’s leadership is now caught between a rock and a hard place. On one hand, the company’s AI push is non-negotiable—it has already spent billions and faces pressure from investors to deliver results. On the other, the morale crisis risks further alienating its remaining workforce. Bosworth’s memo signals an attempt to course-correct, but the damage may already be done. Employees reassigned to AI teams can now reapply for other roles, but the trust has been broken.
Zuckerberg’s admission in an internal memo that “we’ve made mistakes and will almost certainly make more” suggests the company is bracing for further turbulence. The question now is whether Meta can retain enough talent to execute its AI strategy—or if the revolts will continue. With morale at an all-time low and employees calling the work “soul-crushing,” the company’s ability to attract and retain top talent is in serious doubt.
For now, Meta’s AI ambitions remain a work in progress—one that is increasingly seen as a burden rather than an opportunity. The company’s failure to communicate its vision clearly, combined with forced reassignments and menial tasks, has turned its AI unit into a symbol of everything that’s wrong with its current trajectory. Unless leadership can reverse course quickly, the fallout could extend far beyond the Applied AI team.
One thing is clear: Meta’s AI overhaul is not just a technical challenge—it’s a cultural one. And right now, the culture is breaking.
Find more reporting in our Business section.
