Meta will record employees’ keystrokes and use it to train its AI models

In a groundbreaking move, Meta has decided to utilize the interactions of its employees as a new training resource for its artificial intelligence systems. The tech giant aims to analyze mouse movements and keystrokes from its workforce to enhance the capabilities of its AI models. This initiative, initially disclosed by Reuters, underscores the lengths to which technology firms are willing to go in search of training data, which is essential for developing AI systems that can efficiently perform tasks and respond to user inquiries. A Meta representative explained to TechCrunch, "To create AI agents that assist individuals with their daily computer tasks, our models require authentic examples of user behavior. This includes actions like mouse movements, button clicks, and navigating menus. To facilitate this, we are introducing an internal tool designed to record these inputs within specific applications to aid in training our models. We have implemented safeguards to protect sensitive information, ensuring that the data collected is not used for any other purposes." However, this trend raises important questions about privacy in the AI sector. The practice of leveraging internal corporate communications for AI training is becoming increasingly common, with reports highlighting how some startups have had their communication archives mined to serve as data sources for AI development.