Meta pulls back on AI-powered workforce reduction plan

Meta spent much of 2026 testing Project OT, a plan to use AI to automate tasks and reduce headcount by up to 60 percent in some teams, including a second wave of layoffs in November. CEO Mark Zuckerberg ultimately abandoned the plan, though the reasons are unclear. Internal data showed code changes rose 220 percent year over year, but user-facing feature improvements only increased 36 percent, suggesting the AI wasn't effective enough.

Project OT’s internal metrics revealed a stark gap between engineering activity and user-visible results: code changes surged 220 percent year over year, yet feature improvements rose only 36 percent, while technical and security incidents climbed 40 percent. Employee time spent on those incidents grew 70 percent, suggesting the AI-generated code introduced new inefficiencies. Zuckerberg’s July town hall admission—that agentic development had not accelerated as expected—aligned with these figures, and the company’s keystroke and mouse-tracking program further eroded morale, with sentiment scores dropping 19 points.

The plan’s scope was ambitious: executives envisioned shrinking some teams by up to 60 percent, with layoffs, unfilled roles, and performance-based exits. A second November wave was shelved, but Reuters noted that smaller AI-assisted teams persist, and Zuckerberg only ruled out further “company-wide” cuts for the remainder of 2026, leaving open the possibility of targeted reductions or a 2027 expansion.