Meta’s Applied AI unit—roughly 6,500 engineers and product managers drafted into a three-month-old team—has turned toxic. Last week someone hacked a livestreamed, employee-only presentation and launched into a profanity-laced tirade, calling a senior AI executive “a piece of sh*t.” One presenter hid their face; others watched in stunned silence. That meltdown reflects deep frustration over being “draftees” with no real choice but to join or quit, churning out coding puzzles and tasks to train Meta’s AI models. Employees say morale is at rock bottom. Many describe the work as “soul-crushing” and liken it to a gulag. They were told via surprise emails that Meta’s AI still can’t code at human levels and needs real examples from staff. In a leaked audio clip, Mark Zuckerberg explained they’d tap employees because they’re smarter than outside contractors, praising Alexandr Wang, who sold Scale AI to Meta for $14.3 billion and now oversees Superintelligence Labs. The frustration stretches beyond Applied AI. Over 1,600 Meta staffers have signed a petition against monitoring their clicks and keystrokes for training data. On a company-wide call, Chief Product Officer Chris Cox warned employees the environment was “brutal.” Zuckerberg later sent an internal memo admitting mistakes and promising fixes, while reaffirming his goal of making Meta “the best place for the most talented people in the world to make an impact.” Meta declined to comment to TechCrunch.