AI agents meant to replace Meta workers made “large-scale, disruptive actions”

Details

A Reuters investigation, covered by Ars Technica, reveals that Meta launched 'Project OT' (organization transformation) in January 2026, personally directed by Mark Zuckerberg. Internal documents described the goal as making Meta 'AI native' — a company where AI agents handle daily work, workflows are fully automated, and new builds are AI-first. Some teams faced scenario plans with headcount reductions of up to 60%, and one HR executive estimated company-wide cuts could exceed 25%. The most operationally significant detail: AI agents deployed to perform human workers' tasks made 'large-scale, disruptive actions' during the effort — a concrete example of the reliability gap that makes wholesale AI workforce substitution harder than it looks on a whiteboard. The first round of Project OT-linked layoffs did proceed in May 2026; the second round was canceled before final numbers were set. Meta confirmed the scenario planning exercise to Reuters but emphasized that not all scenarios moved forward. The company said thousands of employees were redeployed to newly-established priority teams rather than laid off. Cost savings from the project were partly earmarked to retain and attract high-performing employees — especially those with AI engineering skills — and to fund Meta's broader AI infrastructure buildout. Separately, Meta began selling AI agents to third parties in June 2026, signaling continued commitment to an agent-driven business model despite the internal deployment friction this investigation surfaces.