LUMIA: Linear Probing for Unimodal and MultiModal Membership Inference Attacks Leveraging Internal LLM States

Luis Ibañez-Lissen, Lorena González-Manzano, José María de Fuentes, Nicolas Anciaux, Joaquín García

Abstract: Large Language Models (LLMs) are increasingly used in a variety of applications. Concerns around inferring whether data samples belong to the LLM training dataset have grown in parallel. Previous efforts focus on black-to-grey-box models, thus neglecting the potential benefit from internal LLM information. To address this problem, we propose the use of Linear Probes (LPs) as a method to assess Membership Inference Attacks (MIAs) by examining internal activations of LLMs. Our approach, dubbed LUMIA, applies LPs layer-by-layer to get fine-grained data on the model inner workings. We test this method across several model architectures, sizes and datasets, including unimodal and multimodal tasks. In unimodal MIA, LUMIA achieves an average gain of 14.90% in Area Under the Curve (AUC) over previous techniques. Remarkably, LUMIA reaches AUC > 60% in 65.33% of cases—an increase of 46.80% against the state of the art. Furthermore, our approach reveals key insights, such as the model layers where MIAs are most detectable. In multimodal models, LPs indicate that visual inputs significantly contribute to MIAs—AUC > 60% is reached in 85.90% of the experiments.

External IDs:dblp:conf/esorics/IbanezLissenGFAG25