Meta's Project OT AI replacement experiment collapses amid staff revolt

September 11, 2026 at 04:40 PM2 min

This digest was compiled by AI from multiple sources — links to the originals are below.

Meta's Project OT, an initiative to replace thousands of employees with AI agents, collapsed after staff revolted and the second wave of layoffs was cancelled hours before it began. The company had planned to cut individual teams by up to 60% and sell AI agents to other firms. The reversal came after an Instagram breach through an AI bot and internal resistance.

Key Facts

Meta launched Project OT in January at Mark Zuckerberg's annual offsite in Hawaii, aiming to replace thousands of employees with AI agents.
Internal documents reviewed by Reuters show plans to cut individual teams by up to 60%, with layoffs split into two waves in May and November.
The second wave of layoffs was cancelled hours before it was scheduled to begin, following staff revolt and an Instagram breach through an AI bot.
Meta planned to sell AI agents to other companies for tasks such as meeting scheduling and deal closing.
By June, at least 11 divisions had transitioned to AI-native 'pods' with a 'village' management model of one Org Lead per 30-50 people.

Project OT Launch

Mark Zuckerberg convened top management at his Hawaii residence in January to launch Project OT, or Organization Transformation. Internal documents obtained by Reuters describe the plan: AI agents would take over the daily work of thousands of employees, supervised by small 'talent dense' human teams. Scenario planning included cutting individual teams by up to 60%, with the HR director estimating the scale as comparable to or larger than the 2022-2023 purge that eliminated about 25% of staff. The layoffs were split into two waves, May and November, with a separate initiative to sell AI agents externally for meeting scheduling and deal closing.

AI-Native Philosophy

After ChatGPT's release, Silicon Valley embraced the 'AI native' company philosophy: build the organization around AI from scratch rather than retrofitting existing processes. Meta executives visited Asia and were impressed by startups with AI-centric organizational structures; the head of product said the Singapore office inspired California teams. In July 2025, VP of Product Ime Archibong assembled five pods of two or three engineers plus a designer with AI tools, using four-week sprints instead of six-month planning cycles. An October internal document, the 'AI Native Playbook', scaled the approach: engineer and designer roles disappear, everyone becomes a 'builder', middle management is removed, and agent analysis helps set daily priorities.

Collapse and Reversal

By June, at least 11 divisions had moved to pods, introducing a 'village' model with one Org Lead per 30-50 people and a Pod Lead managing daily work without formal authority. The experiment unraveled: staff revolted, Instagram was breached through an AI bot, and the second wave of layoffs was cancelled hours before it was scheduled to begin. In the end, Meta promised to improve snacks in office kitchens, according to Reuters' reconstruction from dozens of internal documents, posts, and recordings.