PrivDrift: Auditing User-Secret Leakage Under Topic Drift in Active LLM Conversations
A benchmark for testing whether user-disclosed secrets remain recoverable after a conversation moves to unrelated topics.
Natural Language Processing
Codebase RAG
Natural-language exploration of code repositories using syntax-aware parsing and vector retrieval.
LLM Sycophancy and Conformal Risk Control
Studying harmful agreement in language models and uncertainty thresholds designed to reduce sycophantic responses.