Thousands of engineers reassigned, employee morale craters, and a keystroke-tracking program sparks a petition as Zuckerberg's AI-native vision collides with reality
Mark Zuckerberg wanted Meta to become an AI-native company. The company’s own employees had a different term for the experience: a “soul-crushing gulag.”
The great reassignment #
In spring 2026, Meta moved approximately 6,500 to 7,000 engineers into its Applied AI unit, tasking them with training data work. For many of these engineers, people who were hired to build consumer products or infrastructure, the shift felt less like a career opportunity and more like a demotion dressed up in corporate strategy language.
Employee sentiment tells the story in stark numerical terms. Favorable ratings inside Meta dropped from 74% to 55% during the restructuring period. CTO Andrew Bosworth acknowledged that morale had sunk to levels resembling previous company crises.
Then came the layoffs. In May 2026, Meta cut roughly 10% of its workforce, eliminating around 8,000 positions. Those cuts were actually the restrained version: internal plans under what was called Project OT had contemplated reducing headcount by up to 60%, a target that was ultimately abandoned as the backlash intensified.
When surveillance met resistance #
Launched in April 2026, the Model Capability Initiative was designed to track employee keystrokes and clicks, ostensibly to measure productivity and model how AI tools could augment or replace human work. More than 1,600 employees signed a petition against the initiative, calling it a privacy violation.
The productivity paradox #
Zuckerberg himself acknowledged that productivity gains from AI had been disappointing relative to initial expectations. The numbers revealed a telling split: internal code enhancements improved by 220%, but user-facing improvements grew by just 36%.
Meta’s capital expenditure guidance for AI infrastructure reached between $125B to $145B annually. For investors, the 220% internal improvement versus 36% user-facing gain disparity deserves close scrutiny in coming quarters.
Leadership shifts compound the uncertainty #
Yann LeCun, Meta’s head of AI research and one of the most respected figures in the field, departed the company at the end of 2025. Losing a Turing Award winner during an all-in AI pivot is, to put it mildly, not ideal timing.
Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our