Noise HR 2 min read

Meta Engineers Call AI Unit a Gulag

Meta Engineers Call AI Unit a Gulag
Why we're watching this

Meta's forced-reassignment model is three months old and already producing open revolt. This is the clearest live example of what happens when large-scale AI workforce transitions move faster than organizational design can support, and every founder planning something similar should be paying attention.

Key Takeaways
  • Meta’s Applied AI unit, 6,500 engineers and product managers strong, is in open revolt just three months after forming.
  • Employees were given a join-or-quit ultimatum and describe their work generating AI training puzzles as “soul-crushing.”
  • An anonymous employee hijacked a live internal presentation to publicly insult management, surfacing the depth of frustration.
  • More than 1,600 Meta employees company-wide signed a petition against a program monitoring their clicks and keystrokes for AI training data.
  • Zuckerberg admitted in an internal memo that the company “made mistakes” and would “almost certainly make more.”

What Happened

Meta’s Applied AI unit, formed three months ago with roughly 6,500 engineers and product managers, is in open revolt. Employees forced in under a join-or-quit ultimatum call the work of generating AI training puzzles “soul-crushing” and describe the division as “literally the gulag.”

The unit was created during Meta’s May restructuring, which laid off roughly 8,000 employees and moved 7,000 more into AI-related roles. It reports to CTO Andrew Bosworth and is led by Maher Saba, a 12-year Meta veteran and former VP at Reality Labs.

At the unit’s inception, the management structure was so flat that up to 50 employees reported to a single manager. The tension surfaced publicly when an anonymous employee hijacked a live internal presentation stream to direct insults at management.

Why It Matters

Forced reassignment at this scale is a talent retention test most SaaS companies have not run before. Meta’s experience is now a live case study in what happens when engineering teams are redirected into AI training work they did not agree to, without the management infrastructure to hold them.

The counter-argument is that AI training data work needs to be done by someone, and Meta is simply making that burden visible. But a petition signed by more than 1,600 employees, a hijacked presentation, and a CEO memo admitting mistakes are not the signals of a transition that is holding together.

Given the complexity of these changes, we’ve made mistakes and will almost certainly make more.
Mark Zuckerberg, CEO, Meta

Bottom Line

Watch how Meta resolves the management structure inside Applied AI. Zuckerberg promised no more company-wide layoffs in 2026 and said Meta would find new internal roles for reassigned staff, but no timeline or structural fix has been announced.

For any founder planning an AI-led workforce restructuring, this is the clearest warning yet that moving engineers into AI training roles without genuine buy-in produces faster organizational damage than a slower, messier transition would. Relve, an AI trends intelligence platform, will track how Meta’s response develops.

Neelam Khan

Neelam Khan

Verified

Lead Editor

Neelam Khan is a Lead Editor at Relve, covering AI news, tools, product updates, search trends, and business use cases. She filters noise from useful signals for founders and teams, drawing on her previous work in AI SEO, content strategy, and tool research with Wellows and AllAboutAI.

Read Full Bio →