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

  • Luis Ibanez-Lissen
  • , Lorena Gonzalez-Manzano
  • , Jose Maria de Fuentes
  • , Nicolas Anciaux
  • , Joaquin Garcia-Alfaro

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

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.

Original languageEnglish
Title of host publicationComputer Security – ESORICS 2025 - 30th European Symposium on Research in Computer Security, Proceedings
EditorsVincent Nicomette, Abdelmalek Benzekri, Nora Boulahia-Cuppens, Jaideep Vaidya
PublisherSpringer Science and Business Media Deutschland GmbH
Pages186-206
Number of pages21
ISBN (Print)9783032078834
DOIs
Publication statusPublished - 1 Jan 2026
Event30th European Symposium on Research in Computer Security, ESORICS 2025 - Toulouse, France
Duration: 22 Sept 202524 Sept 2025

Publication series

NameLecture Notes in Computer Science
Volume16053 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference30th European Symposium on Research in Computer Security, ESORICS 2025
Country/TerritoryFrance
CityToulouse
Period22/09/2524/09/25

Keywords

  • Large Language Models
  • Large Multimodal Models
  • Linear Probes
  • Membership Inference Attacks

Fingerprint

Dive into the research topics of 'LUMIA: Linear Probing for Unimodal and MultiModal Membership Inference Attacks Leveraging Internal LLM States'. Together they form a unique fingerprint.

Cite this