Meta's AI Unit in Crisis: Internal Turmoil Reveals Challenges with Rapid Restructuring
Key Takeaways
- ▸Meta's 6,500-person Applied AI team faces a morale crisis, with employees describing work assignments as soul-crushing and menial, including puzzle-generation tasks unrelated to their previous software development roles
- ▸A company-wide AI restructuring that eliminated 8,000 positions (10% of workforce) has created stress and low morale across multiple departments, with employees reporting that the pace and brutality of change is unsustainable
- ▸Over 1,600 Meta employees signed a petition opposing a keystroke-monitoring program designed to collect AI training data, reflecting broader concerns about privacy and employee welfare during the transition
Summary
Meta's newly formed Applied AI team, created in March with approximately 6,500 engineers and product managers to support the company's AI research efforts, is experiencing widespread employee dissatisfaction and crisis-level morale issues. According to WIRED reporting based on employee accounts and internal recordings, workers describe their assignments as soul-crushing and menial—including puzzle generation tasks to test AI models—with one employee characterizing the unit as a 'gulag.' An expletive-filled outburst during a livestreamed employee meeting highlighted the frustration, with workers reporting they feel disconnected from meaningful work and lacking purpose.
The turmoil within Applied AI reflects broader organizational challenges following Meta's major AI-focused restructuring last month, which resulted in the layoff of 8,000 employees (10% of the workforce). Multiple divisions including data center engineering and Instagram are experiencing elevated stress and low morale. Adding to the workplace tensions, more than 1,600 employees signed a petition against Meta's initiative to monitor employee keystrokes and clicks to generate AI training data, forcing the company to make limited concessions including the ability to pause monitoring for up to 30 minutes.
Meta's leadership has acknowledged the dysfunction. Chief Product Officer Chris Cox described the environment as 'brutal' and 'difficult,' while CEO Mark Zuckerberg admitted in a Friday internal memo that the company has made mistakes during the reorganization. Cox stated the company needs to recalibrate its relationship with AI, calling it 'neither god, nor the devil.' These admissions suggest Meta recognizes that its aggressive restructuring may be undermining the very talent and engagement needed to succeed in AI development.
- Meta's CEO and leadership team have publicly acknowledged organizational missteps and a difficult environment, signaling awareness of serious retention and engagement risks in the company's AI initiatives
Editorial Opinion
Meta's crisis within its Applied AI unit reveals a critical flaw in the company's AI strategy: rapid reorganization without clear purpose or employee engagement is likely to undermine long-term innovation. By treating talented engineers as interchangeable labor units and assigning soul-crushing work, Meta risks losing exactly the talent it needs to compete in AI. The keystroke monitoring and the dismissive characterization of work suggests a company more focused on extraction than collaboration—a recipe for failure in talent-dependent AI development.
