PA-HDP: Dynamic Privacy Protection for RAG Under Query-Driven Risks

Jul 17, 2026

A new framework, PA-HDP, is proposed to address privacy risks in retrieval-augmented generation (RAG) systems by recognizing that privacy leakage is dynamic and depends on the user's query. PA-HDP uses a prompt-aware risk hierarchy and adaptive protection mechanisms to assess and mitigate privacy risks on a per-query basis. Experimental results show that PA-HDP reduces privacy leakage and maintains retrieval quality better than previous static, document-level approaches.

Why it matters: This work introduces a more nuanced and effective approach to privacy in RAG systems by adapting protection to the actual sensitivity of content in response to specific user queries.

Full story at: arXiv Cryptography and Security ↗