Mark Zuckerberg's shield — Meta CTO Andrew Bosworth — is doubling down on an AI transformation that's tearing the company apart from the inside. On the same day Meta laid off 8,000 people in a surprise purge, sending termination emails at 4 AM, the company forced the Model Capability Initiative (MCI) monitoring tool onto the machines of remaining employees. It logs every keystroke, mouse click, and screenshot. The result? Furious staffers have started passing out flyers around the office calling Meta an "employee data extraction factory" and gathered over 1,500 signatures on a protest petition.
The core grievance: the company is using their daily work to train the very AI systems that will eventually replace them. On internal forums, employees asked how to turn off the monitoring. Bosworth's reply was cold: no one has the option to opt out of data collection on company computers. His response in internal chats got flooded with crying, shocked, and angry emoji reactions.
Zuckerberg has publicly tried to soothe concerns, saying the data is for training, not performance reviews. But Bosworth's internal memo, the "Agent Transformation Accelerator," paints a different picture. He flat out states Meta's vision: most future work will be done by AI agents, and humans will be there to guide, review, and help improve them.
At an all-hands meeting in late April, Zuckerberg defended MCI by claiming Meta engineers' average intelligence is way above outsourced contractors — so their elite workflow patterns are uniquely valuable training data. But after the May 20 mass layoffs, staffers under stack-ranking pressure, mortgage stress, and H1B visa anxiety realized that high performance doesn't buy safety. They're being used to inflate the stock price — and drained for every last bit of data value before they're shown the door.