Meta plans to use its employees’ mouse movements and keystrokes as a new source of training data for its artificial intelligence models. The initiative aims to improve the company’s AI capabilities by utilizing actual user interactions.
This development underscores the intensifying competition among tech firms to secure training data, which is crucial for enhancing the effectiveness of AI systems. According to Reuters, tech companies are increasingly exploring innovative methods to gather this vital information.
A Meta spokesperson stated, “If we’re building agents to help people complete everyday tasks using computers, our models need real examples of how people actually use them.” The spokesperson added that the initiative involves launching an internal tool designed to capture user inputs in various applications.
The tool will track mouse movements, button clicks, and navigation patterns to provide data for training. Meta asserts that it has implemented safeguards to protect sensitive content, ensuring that the data collected will not serve any other purpose beyond model training.
However, this approach raises privacy concerns within the AI industry. Reports indicate that other tech startups have been targeted for their internal communications, such as Slack messages and Jira tickets, which are then converted into AI training data.
