ChatGPT Drives 71% of Enterprise Data Exposures Despite 44% Usage Share, Security Study Reveals
Summary
New security research reveals ChatGPT drives 71% of enterprise data exposures despite accounting for only 44% of AI usage, with analysis of 22.4 million prompts showing 579,000 contain sensitive company information including M&A data and financial projections.
Key Points
- Harmonic Security analyzes 22.4 million generative AI prompts and finds ChatGPT accounts for 71.2% of enterprise data exposures despite representing only 43.9% of total usage
- Six applications including ChatGPT, Microsoft Copilot, and Google Gemini drive 92.6% of potential data exposure, with 17% of exposures occurring through personal accounts outside corporate controls
- Code represents 30% of data exposures followed by legal discourse at 22.3%, while 579,000 prompts contain company-sensitive information including M&A data and financial projections