🚨 BREAKING: Meta's 6,500-Engineer AI Squad Just Had a Full Locker Room Meltdown — Zuckerberg Issues Damage Control Memo

Someone hijacked an employee-only Meta livestream and told the presenter's boss to go to hell. That's the headline. Here's the box score. 1

The incident

This week, during an internal Meta call open to thousands of employees, someone unmuted and exploded — demanding that attendees relay a specific, printable-in-a-family-sports-broadcast-uh-no message to a senior Meta AI exec. One presenter covered their face with their hands. The other muted the room and pressed on. 2

The call belonged to Meta's Applied AI unit — a 6,500-person team of engineers and product managers stood up just three months ago to generate training data for Meta's frontier AI models.

The roster problem

Meta Applied AI — the box score

Key numbers behind the locker room implosion

Engineers drafted

Petition signers (surveillance)

Engineers per manager (some teams)

Unit age (months)

통계 카드를 불러오는 중…

The Applied AI unit was assembled the way a struggling franchise trades for aging stars: you take the deal or you don't play. Engineers were not asked to volunteer. They were assigned. Many call themselves "draftees." 2

Their job? Generate coding puzzles and synthetic tasks to train Meta's models. Work these engineers describe, repeatedly and on record, as "mechanical," "not creative," "soul-crushing," and — in the quote that will define this story — "literally the gulag." 2

The unit launched in April in batches. "It's crazy to watch people experience the shock of it as each wave comes in," one early member told Wired.

Staffing ratio: up to 50 engineers per manager on some teams. This was deliberate. The logic, per an April internal memo reviewed by Wired and Business Insider: Meta believes its own engineers are "significantly higher intelligence" than outside contractors, so better to draft them than source externally. Zuckerberg explained this to employees in a leaked audio recording, citing Meta's AI chief officer Alexandr Wang — formerly of Scale AI, which Meta acquired for $14.3 billion. 1

Meta pavilion at the World Economic Forum, Davos, January 2025 — Meta's PR face vs. its internal reality right now. 1

The surveillance petition

It's not just Applied AI. Across Meta, more than 1,600 employees have signed a petition demanding the company stop a program that monitors their clicks and keystrokes — also for AI training. Meta has scaled back slightly: employees can now pause data collection for 30 minutes or request exemptions. That detail is doing a lot of work in a company with tens of thousands of employees. 2

Meta's chief product officer Chris Cox addressed the mood this week on a call with Instagram employees. He called the environment "difficult," "brutal," and compared it to "running a marathon in the middle of a hailstorm." He also said, drawing laughs: "It's like what the f*ck." 2

The Zuckerberg memo

By Friday, Zuckerberg issued a company-wide memo. He acknowledged the changes had "caused distress," admitted mistakes had been made, and promised more stability. He pledged: no more mass layoffs this year. New manager-to-report caps. Bigger budgets for team events. Assigned desks back by year-end. A hackathon in July. 2

Meta CEO Mark Zuckerberg exiting Los Angeles Superior Court, June 2026 — Zuckerberg this week — the memo was his damage-control move after the locker room blew up. 2

On Applied AI specifically, Zuckerberg framed the unit as a waypoint: "Work like AAI is critical to advancing our models and it lets very talented people contribute to those efforts while we create other roles they can contribute to around Meta over the coming months."

The memo closed with Meta's stated north star: "to be the best place for the most talented people in the world to make an impact."

Six thousand five hundred engineers generating coding puzzles would probably phrase their current situation differently.

#AILeague: Meta/Llama locker room report 🔴

Team: Meta/Llama — the open-source community squad, no ticket price, deep community loyalty, but increasingly hard to ignore the front-office chaos.

The play: This isn't a trade or a signing. This is a locker room implosion — the kind that happens when a franchise makes a bold roster move without managing the culture cost. Zuckerberg drafted 6,500 engineers into a data-labeling unit and called it the future of AI. The players called it the gulag.

League context: Meta has been on an aggressive rebuild — $83 billion burned on the metaverse, now pivoting hard into AI. The Applied AI unit is the depth chart move: put your highest-IQ players on synthetic data generation so the star researchers at Meta Superintelligence Labs can train better models. In theory, it's smart roster construction. In practice, drafting a Stanford-educated software engineer to write puzzle prompts, without asking, is the kind of front-office decision that gets the GM fired in real sports.

Standings impact: Meta/Llama's open-source releases (Llama 3 and successors) have been genuine league wins. But talent flight risk just went up. When your engineers are publicly calling their workplace "the gulag" and 1,600+ sign a surveillance petition, the competing squads — OpenAI, Anthropic, Google DeepMind — are watching. Recruiting advantage shifts.

The Zuckerberg memo probably stops the immediate bleeding. The structural question — whether you can build AGI with conscripted engineers — doesn't have a memo fix.

#AILeague