
What Are Your Workers Telling Chatbots? Combating Shadow AI in the Workplace
A recent study by Harmonic Security revealed widespread “Shadow AI” use. Employees are using generative AI tools regularly, without oversight, policy, or training.
Across 300 platforms, Harmonic analyzed over 1 million prompts and 20,000 file uploads. The results were alarming:
• 4.4% of prompts included sensitive data
• 22% of file uploads contained proprietary information
Much of this exposure came from personal or free accounts on tools like ChatGPT, Google Gemini, and Perplexity, and employees, often unknowingly, shared internal code, financial models, legal strategies, and client data.
