Meta's AI push strains staff: 10% layoffs, surveillance spark revolt

What happened

Meta has cut 10 percent of its workforce—roughly 8,000 jobs—this spring and reassigned 7,000 employees since early 2025. The company rolled out a controversial AI training program called the Model Capability Initiative in April that captured clicks, keystrokes, and browsing activity of US employees to train AI agents (software capable of performing tasks independently). The program was suspended on June 22 after more than 1,600 employees signed a petition opposing it.

Why it matters

Meta is spending up to $145 billion（約23兆円） on AI investments this year, nearly twice last year's figure, to compete in the AI race. Employees have complained of being assigned "mind-numbing" tasks to train machines or of automating away their own jobs. The internal climate has grown heated as the company imposes sweeping cuts and increased monitoring in the name of efficiency—a sharp contrast to Meta's strong finances, where net income rose to more than $26 billion（約4.2兆円） in the first three months of 2026.

What to watch

The Model Capability Initiative paused after private conversations and performance data inadvertently became accessible to all staff, raising European regulatory concerns since it captured exchanges between employees on both continents. Zuckerberg, who has near-absolute power over Meta, has made AI the company's North Star and defended the program by arguing "AI models learn by watching really smart people do things."