This research addresses critical privacy vulnerabilities in Large Language Models that can expose sensitive personal information across various sectors including healthcare and finance.

• Identifies risks of unintentional PII disclosure when LLMs are trained on diverse datasets
• Examines techniques to prevent data extraction attacks that compromise user privacy
• Proposes architectural frameworks for privacy-by-design in generative AI applications
• Highlights the importance of balancing utility with robust privacy protection measures

This work is particularly significant for security professionals as it provides actionable strategies to implement privacy guardrails in an era where LLMs are increasingly deployed in sensitive environments handling confidential information.

Privacy Preservation in Gen AI Applications