Zuckerberg’s 4 AM Layoff of 8,000 Staff: Using Employees’ Keyboard Data to Train AI

In the third round of workforce reductions announced for 2026, Meta dismissed about 8,000 employees, representing roughly 20% of its global staff. The termination notices were sent at 4 am Singapore time, and access to laptops, corporate email, and internal systems was revoked instantly.

The layoff operation was internally codenamed “Pralaydin,” meaning “Day of Destruction.” Prior to the cuts, more than 1,000 employees signed a petition protesting a covert monitoring program that records keyboard strokes, mouse movements, and screenshots to train the company’s AI models.

Despite the protest, the program continued. A leaked recording from the April 30 all‑hands meeting captured Mark Zuckerberg admitting that Meta uses employees’ work to teach AI, stating that the company’s engineers are “much smarter than the average person you could hire for a task.”

Evercore estimates that eliminating the 8,000 positions saves Meta approximately $3 billion per year. The article also notes that Zuckerberg wrote a $200 million check to recruit Apple executive Peng Peng, pledged $1.5 billion over six years to retain AI researcher Andrew Tulloch, and spent $14.3 billion to bring Alexandr Wang on board.

Financially, Meta reported Q1 2026 revenue of $56 billion and net profit of $27 billion, a 33% year‑over‑year increase. The company’s primary concern is allocating its roughly $130 billion budget toward AI data centers and Nvidia GPUs.

For the 2,000 “survivors,” Meta created an “Applied AI and Engineering” division led by VP Maher Saba, using the collected employee data for re‑organization. Severance packages include 16 weeks of base salary, an additional two weeks per year of service, and 1.5 years of health coverage, with an average payout of $360 k.

Former engineers describe a bleak hiring market: despite generous severance, many struggle to secure new roles, and applications to hundreds of companies receive no response. Internal forums saw employees sharing “Salad” emojis as a grim salute, while some, like ex‑engineer Jeremy Bernier, left the company abruptly for travel abroad.

The piece concludes by linking the layoffs to broader industry debates, referencing Andrew Ng’s criticism of “AI apocalypse” narratives as a business tactic, and invites readers to consider the implications of using employee labor as data for AI development.