Meta Applied AI engineers revolt over surveillance

Meta's aggressive reassignment of software engineers to manual data-labeling tasks reveals a massive bottleneck in sourcing high-quality training data for frontier models.

• –Reassigning high-caliber software developers to repetitive data labeling is an expensive, low-leverage move that highlights the industry's data desperation.
• –Implementing keystroke surveillance to monitor these engineers degrades workplace trust and accelerates developer burnout and attrition.
• –This friction underscores a growing industry-wide shift where traditional software engineering roles are displaced by generative AI data pipeline tasks.