MRMMIA is a multi-recall-probe membership inference attack that extracts signals from chat agent memory and outperforms baselines in black-, gray-, and white-box settings.
Mask-based membership inference attacks for retrieval- augmented generation, 2025
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CR 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
MRMMIA: Membership Inference Attacks on Memory in Chat Agents
MRMMIA is a multi-recall-probe membership inference attack that extracts signals from chat agent memory and outperforms baselines in black-, gray-, and white-box settings.