Meta scraps Project OT after AI agents trigger surge in technical incidents
The social media giant has cancelled a restructuring plan that sought to cut team headcounts by up to 60 percent, citing a 40 percent rise in major technical and security incidents caused by AI agents.
Meta has cancelled Project OT, an internal initiative designed to transform the company into an “AI native” organisation by reducing some team headcounts by as much as 60 percent. The plan, codenamed for “organisation transformation,” was created in January 2026 and involved two rounds of layoffs. While the first round of cuts occurred in May, CEO Mark Zuckerberg cancelled the second wave scheduled for November, according to a report by Reuters.
Internal documents reviewed by Reuters indicate that the AI agents intended to replace human workers made “large-scale, disruptive actions,” leading to a 40 percent increase in major technical and security incidents compared to the prior year. Consequently, employee time spent resolving these AI-related problems increased by as much as 70 percent. Meta stated it did not move forward with every scenario from the exercise, which aimed to automate workflows and sell AI agents to third parties.
The initiative was part of a broader effort to make the company “AI native,” a term defined in internal planning documents as a state where AI-ready tools interact, workflows are automated, and new builds are AI-first. Meta began selling AI agents to third parties in June 2026, aligning with this definition. However, the company also paused a program that tracked employees’ keyboard and mouse input to train these agents, a move that reportedly hurt morale.
Productivity gains from the increased use of AI have been mixed. Code changes to internal software platforms and infrastructure were up 220 percent year-over-year, according to a post by Meta CTO Andrew Bosworth in early June. In contrast, changes leading to new or upgraded features reaching Meta users were only up 36 percent. This discrepancy highlighted uncertainty about whether increased AI use would boost overall productivity.
Zuckerberg reportedly acknowledged in July that the “trajectory of the agentic development over at least the last four months hasn’t really accelerated in the way that we expected.” The CEO had initially set Project OT in motion and directed executives to proceed with changes to management structures. Meta confirmed to Reuters that the plan explored scenarios for reducing team headcounts by 60 percent, though it declined to confirm which specific teams were affected.
One HR executive reportedly estimated that Meta would have reduced its headcount by about 25 percent or more under these scenarios. Meta told Reuters that it cancelled Project OT before determining the final number of layoffs. The company also noted that it had used some of the money saved from layoffs to pay for high-performing employees, especially those with AI engineering skills.
The cancellation of Project OT underscores the challenges organisations face when analysing the best uses for AI and determining when the technology is a better fit for certain tasks than employees. Meta’s experience suggests that even some of the most eager organisations may struggle to replace various human workloads with AI while managing the risks of overzealous implementation.
