This research introduces a privacy preservation pipeline that protects sensitive information when users interact with cloud-based large language models.

• Addresses critical risks of data breaches and unauthorized access to personal information
• Enables secure LLM interactions while maintaining data privacy
• Particularly valuable for applications handling sensitive user data, including medical information
• Balances privacy protection with maintaining LLM functionality

As organizations increasingly integrate cloud LLMs into their services, this framework provides a vital security layer that helps comply with data protection regulations while maintaining user trust.

PRIV-QA: Privacy-Preserving Question Answering for Cloud Large Language Models