Meta Project OT Weighed 60% Team Cuts as AI Incidents Rose 40%

Summary

In January 2026, Meta executives created Project OT, short for organization transformation, after CEO Mark Zuckerberg directed management changes intended to make Meta AI native. Scenarios assigned AI agents much of the daily work performed by thousands, supervised by small human teams. Some teams could shrink 60%, while one HR executive estimated headcount cuts of about 25% or more. The plan combined redeployments, open role closures and two layoff rounds, with some savings funding higher pay for top performers, especially AI engineers. May layoffs proceeded and thousands moved to new priority teams, but Zuckerberg immediately canceled the planned November wave. As of August 26, Meta had not identified affected teams and said OT ended before a final layoff total was set.

Meta also piloted smaller engineering, research and at least eight other teams after the previous October’s AI-Native Playbook proposed fewer middle managers and agent assisted task prioritization. HR employees and AI systems supported promotion reviews, although Meta says people retained final decisions. Its vision included automated workflows, AI first development and third party agent sales, which began in June.

Results weakened the case for deeper cuts. March and April layoff warnings and keyboard and mouse monitoring to train agents, since paused, hurt morale. In early June, CTO Andrew Bosworth said internal platform and infrastructure code changes rose 220% year over year, but new or upgraded features reaching users increased only 36%. Internal posts linked agents to large scale disruptive actions, a 40% increase in major technical and security incidents, and up to 70% more employee time resolving them; Meta declined comment. In July, Zuckerberg said agentic development had not accelerated as expected over at least four months, while his reason for reversing course remained unclear.