Meta Halts Plan to Replace 60% of Staff With AI

Meta recently halted an internal initiative known as Project OT, or Organization Transformation, which aimed to make the company "AI native" by replacing up to 60 percent of employees on certain teams with artificial intelligence. According to a Reuters investigation, executives abandoned the plan at the last minute after internal metrics revealed that the automation push was causing severe operational issues rather than boosting actual productivity.

Data shared by Meta Chief Technology Officer Andrew Bosworth in June showed that while AI-driven code changes to internal platforms surged 220 percent year-over-year, actual feature upgrades reaching users only increased by 36 percent. Meanwhile, unchecked AI agents triggered a 40 percent spike in major technical and security incidents, including service disruptions and potential data leaks. Consequently, the time engineers spent resolving these emergencies jumped by 70 percent. Additionally, an April mandate requiring tracking software on U.S. employees' devices to record keystrokes and clicks for AI training sparked widespread worker backlash, which Meta tried to soothe with promises of better office snacks and travel budgets.

For enterprise IT leaders and software practitioners, Meta's setback serves as a stark warning against treating raw AI output as a proxy for business value. Industry analysts point out that AI can easily manufacture activity at machine speed without generating real-world utility. As Acceligence CEO Justin Greis noted, massive increases in code output often just mean ten times more material that human staff must validate, secure, and maintain. Practitioners must establish rigorous guardrails, proving an AI agent's capabilities in production before granting it autonomy or reducing human oversight.