Meta's AI "Gulag" just exploded. 6,500 engineers in open revolt.

6,500 Meta engineers just called their own AI unit "the gulag."

Not a leak. Not a rumor. A hijacked internal livestream where an employee demanded the audience call a senior AI executive "a piece of sh*t."

Here is what happened.

In May, Meta cut 8,000 jobs and forced 7,000 engineers into a new Applied AI unit via surprise email. No choice. Join or quit.

Their job: generate puzzles and coding problems to train Meta's AI models. Soul-crushing work for people hired to build frontier systems.

Zuckerberg defended it in a leaked recording. He said Meta employees have "significantly higher" intelligence than outside contractors. The people doing the work heard that.

1,600 employees across Meta signed a petition against keystroke and click monitoring for AI training data. CPO Chris Cox called the environment "brutal" on an internal call.

Friday, Zuckerberg sent a memo. He admitted the changes "causeddistress" and that Meta "made mistakes." He promised no more company-wide layoffs in 2026. He offered hackathons and assigned desks.

None of that fixes the structural problem.

Meta paid $14.3 billion for Scale AI and hired Alexandr Wang to run its superintelligence lab. What 6,500 engineers actually got was the data-labeling operation repackaged as "agent training." The compensation packages were frontier-research-grade. The work is operational data generation.

That gap is not a morale issue. It is a category error in org design.

If you are restructuring your workforce around AI, audit the assignment logic. The revolt at Meta is not about culture. It is about mismatched expectations between what people were told they would do and what they are actually doing.

The same pattern is forming inside every enterprise that forces engineers into AI roles without redesigning the work itself.