Meta AI Department Engineer Uses Profanity During Internal Live Stream Meeting
An engineer from Meta's AI department used profanity during an internal live stream meeting attended by thousands, complaining about being a "company dog" and insulting executives, causing the speaker to awkwardly cover their face.
The department was established in March this year, with about 6,500 engineers being forcibly transferred, given only the options to accept or resign, and forced to engage in tedious labeling work under strict monitoring for generative AI training puzzles. Employees mockingly referred to it as a "Gulag" concentration camp and initiated a petition with 1,600 signatures in protest.
In a memo on June 12, Zuckerberg acknowledged the harsh environment, promising to limit the number of subordinates per manager, avoid large layoffs, and to treat the department as a temporary transfer station to reassign employees to more valuable positions.
Source: Public Information
ABAB AI Insight
Meta previously, under Alexandr Wang's push, massively requisitioned high-paid engineers for data labeling, having briefly halted similar practices after acquiring Scale AI, but revived and expanded this model in March 2026, leading to paralysis in other departments like the security team due to personnel reallocations. Historically, Zuckerberg has repeatedly addressed AI resource allocation conflicts through internal restructuring.
The company is shifting capital from chaotic internal manpower consumption to structured budget control, while heavily investing in data centers and model infrastructure. Zuckerberg's memo serves as a calming signal to stabilize core talent and maintain the pace of AI expansion.
Similar dissatisfaction among engineers over repetitive tasks in early AI projects at Google and Microsoft led to attrition; Meta is currently transitioning from AI infrastructure expansion to internal talent efficiency control.
Essentially, this reflects an acceleration of technological substitution under capital concentration: the high demand for AI training forces companies to shift from external outsourcing to internal high-IQ labor, but rapid scaling exposes management friction, driving a shift from forced requisition to job optimization and restructuring to maintain talent pricing power and long-term innovation output.
ABAB News · Cognitive Law
The more coercion, the stronger the resistance; the earlier the empowerment, the higher the loyalty.
High salaries buy time, inefficiency is waste; improper structure turns talent into consumables.
Short-term requisition fills hunger, long-term optimization builds barriers; the moment of loss of control is the starting point for reconstruction.