Li Xiong | Security Research Group

Beyond Indistinguishability

11 May 2026 LLM, memorization, David Evans, Ruixuan Liu, Li Xiong

Ruixuan Liu has made an interactive demo of our work on measuring extraction risk in LLMs:

Ruixuan will present the paper at IEEE Security and Privacy:

Ruixuan Liu, David Evans, Li Xiong. Beyond Indistinguishability: Measuring Extraction Risk in LLM APIs. In 47^th IEEE Symposium on Security and Privacy (Oakland). [arXiv] [Code]

All Posts by Category or Tags.