Sources: Meta internally exposed data from its employee-tracking program meant to help train its AI models, including full prompts and private conversations (Wired)

Meta's employee-tracking program designed to collect keystroke data for AI model training suffered an internal data exposure, compromising sensitive employee information including full prompts and private conversations. This incident underscores critical risks in data collection initiatives and highlights the tension between leveraging employee data for AI advancement and protecting workforce privacy, requiring IT leaders to reassess their own monitoring and data governance practices. The exposure demonstrates that even well-intentioned AI training programs can create significant security, compliance, and employee trust vulnerabilities if not properly secured and transparently managed.