Meta has introduced a new internal tool that tracks mouse movements, keystrokes, and button clicks to generate data for training its artificial intelligence models. The system converts user interactions into structured datasets without requiring explicit employee consent for each recorded action. Company representatives confirmed the tool operates across Meta’s internal platforms, including internal dashboards and development environments.
The initiative follows Meta’s broader push to expand its AI capabilities by leveraging real-world usage patterns. Executives stated the data helps refine models used in products like Meta AI Assistant and content moderation systems. The company did not disclose how long the collected data is retained or whether employees can opt out of the program.
Privacy advocates raised concerns about the scope of data collection and its compliance with existing workplace regulations. A Meta spokesperson said the tool adheres to internal privacy policies and undergoes regular audits. However, labor unions and digital rights groups have criticized the practice as an invasion of worker privacy.
Meta’s move mirrors similar strategies adopted by other tech giants, including Google and Microsoft, which have also integrated employee activity into AI training pipelines. These efforts aim to improve model accuracy but have sparked debates over workplace surveillance and data ownership.
The company plans to integrate the tool more deeply into its internal workflows over the coming months. No timeline has been provided for external audits or third-party reviews of the data collection process.
Source: techcrunch.com