Meta's months-old AI unit is a soul-crushing gulag, say the engineers stuck inside it

Internal Turmoil Plagues Meta's New AI Unit: Engineers Reportedly Facing "Soul-Crushing Gulag"

A new report from Wired details significant internal unrest within Meta's recently established Applied AI unit, a division formed just three months ago with approximately 6,500 engineers and product managers. The report suggests that the unit is on the "verge of revolt," with employees describing their work environment in stark terms, including calling it a "soul-crushing gulag." This revelation comes as Meta aggressively pivots its strategic focus towards artificial intelligence, aiming to become a leader in the rapidly evolving AI landscape.

Meta's Ambitious AI Pivot and Workforce Restructuring

Meta has been making substantial investments and organizational changes to accelerate its AI initiatives. In August 2025, the company announced a major overhaul of its AI division, including downsizing some departments and establishing Meta Superintelligence Labs to accelerate AI development. This restructuring aimed to foster innovation and efficiency by creating specialized teams. Reports from May 2026 indicated that nearly 8,000 employees could be affected by layoffs, with around 7,000 reassigned to new AI-focused teams. This aggressive push reflects Meta's determination to compete with rivals like Google and OpenAI. The Applied AI unit, at the heart of the recent controversy, was created in March 2026 to support researchers at Meta Superintelligence Labs. It is headed by Maher Saba, who previously served as a vice president in Meta's Reality Labs division, and reports to Meta CTO Andrew Bosworth. The unit's primary responsibility involves creating "the data engine that helps our models get better, faster," according to an internal memo.

The Core of the Discontent: Forced Transfers, Monotonous Work, and High Ratios

The Wired report highlights several key factors contributing to the severe employee dissatisfaction within the Applied AI unit:

• Forced Transfers: Many employees in the Applied AI unit claim they were given no choice but to join the new division or resign, leading some to label themselves "draftees." This mandatory reassignment stands in contrast to the typical voluntary nature of highly skilled tech roles.
• "Soul-Crushing" Work: The duties assigned to engineers in the unit reportedly involve tasks such as generating puzzles and coding challenges to train AI systems. This work is described by employees as "menial," "demoralizing," and lacking the creativity and technical challenges they expected from a company like Meta. One employee explicitly stated, "most people find the work soul-crushing," while another called it "literally the gulag."
• High Manager-to-Employee Ratio: The unit reportedly adopted a flat organizational structure with an unusually high ratio of up to 50 individual contributors per manager. This ratio is significantly higher than the commonly accepted limit for effective span of control in management, raising concerns about bottlenecks in communication and decision-making.

Public Protests and Surveillance Concerns

The internal unrest escalated to a notable incident during a livestreamed, employee-only meeting. An individual reportedly interrupted the call, which was attended by thousands, with a "vulgar tirade" urging participants to tell a senior Meta AI executive that he was a "piece of sh*t." This public display of defiance underscores the deep-seated frustration among some employees. Adding to the discontent is a controversial company initiative that tracks employees' mouse clicks and keystrokes for AI training data. Over 1,600 Meta employees company-wide signed a petition opposing this system, with some describing it as "very dystopian" and fearing it was "just the latest way they're shoving AI down everyone's throat." Although Meta has since scaled back parts of this program, allowing employees to pause data collection and request exemptions, the initial backlash highlights a significant trust issue within the company.

Leadership Acknowledges "Mistakes"

In response to the growing internal turmoil, Meta CEO Mark Zuckerberg reportedly addressed the situation in an internal memo. He acknowledged that the company had made "mistakes" in its AI workforce transformation and conceded that recent changes had caused distress. Zuckerberg reportedly pledged that Meta would not conduct additional mass layoffs this year and outlined intentions to improve morale, including increased funding for team activities and a company-wide hackathon. He also indicated plans to scale back the practice of widening manager oversight responsibilities. Meta's Chief Product Officer, Chris Cox, also reportedly called the current atmosphere "brutal" during an internal call.

Broader Implications for Meta's AI Ambitions

This widespread dissatisfaction within a critical AI unit poses significant challenges for Meta's ambitious AI strategy. The company is investing heavily, with projected capital expenditures for 2026 ranging from $125 billion to $145 billion, nearly double its 2025 outlay, primarily for AI infrastructure and development. Such internal friction could hinder Meta's ability to attract and retain top AI talent, which is crucial in the highly competitive AI industry. Earlier reports from July 2025 already pointed to a "dysfunctional" AI unit characterized by "fear and confusion," with many workers feeling disengaged. The departure of key figures like Yann LeCun, Meta's former Chief AI Scientist, in November 2025, also highlighted internal tensions and a shift in Meta's AI strategy towards faster product rollout over long-term research. Maintaining employee morale and fostering a productive work environment are paramount for innovation, especially in a field as dynamic as AI. The current situation suggests that Meta faces a significant internal challenge in balancing its aggressive AI goals with the well-being and engagement of its workforce.

Frequently Asked Questions

What is the "Applied AI unit" at Meta?

The Applied AI unit is a division within Meta, established in March 2026, comprising approximately 6,500 engineers and product managers. Its purpose is to support Meta Superintelligence Labs researchers by creating data and challenges to train and evaluate AI systems.

Why are employees in Meta's Applied AI unit reportedly unhappy?

Employees in the unit are reportedly unhappy due to several reasons, including being forced into the unit, finding the work "soul-crushing" and "menial" (like creating puzzles for AI training), a high manager-to-employee ratio (up to 50:1), and concerns over a controversial system that tracks mouse clicks and keystrokes for AI training data.

How has Meta's leadership responded to the employee dissatisfaction?

Meta CEO Mark Zuckerberg has acknowledged that the company made "mistakes" during its AI workforce transformation and that recent changes caused distress. He has pledged to address these issues by not conducting further mass layoffs in 2026, increasing budgets for team activities, organizing a hackathon, and planning to scale back the high manager oversight ratios.

What are the broader implications of this internal unrest for Meta's AI strategy?

The internal unrest could hinder Meta's ability to attract and retain top AI talent, which is crucial for its ambitious AI development goals. It also highlights the challenges of rapidly restructuring a large workforce to pivot towards new strategic priorities, potentially impacting innovation and overall productivity in the highly competitive AI sector.