VLALeaks: Membership Inference Attacks against Vision-Language-Action Models
On the map: Vision-language-action models → Safety and adversarial attacks
also in 1 more
Abstract (arXiv)
Vision-Language-Action (VLA) models enable end-to-end robot control and have garnered widespread attention. However, the memorization of training data inherent to VLA, coupled with the high cost of robotic data acquisition, raises serious concerns regarding data privacy leakage and intellectual property infringement. Membership inference attacks (MIAs) aim to determine whether a given sample belongs to the training set. While representing a significant privacy threat, this attack remains underexplored in the context of VLA models. To bridge this gap, we propose VLALeaks, which is based on attention discrepancies in VLA models. We reveal, for the first time, the privacy vulnerabilities of VLA models. Specifically, it comprises a two-stage process: (1) membership feature extraction, and (2) attack model construction. Experimental results across multiple VLA benchmarks demonstrate that VLALeaks readily reveals membership information and achieves optimal attack AUC and TPR@1%FPR, highlighting the privacy vulnerabilities in current VLA model deployments. Our work is the first systematic study of MIAs on VLA models, aiming to provide insights for secure and trustworthy VLA models.
Key facts. Cited by 2 papers in the map (data through 14 Oct 2026). Placed in Safety and adversarial attacks, one of 13 directions in Vision-language-action models. Builds on 38 of its 56 references that are in the map. First posted on arXiv 13 Jun 2026.
Builds on
38 of 56 references.
Closest ancestors · 3
Same direction, newest first
- 2026When Attention Betrays: Erasing Backdoor Attacks in Robotic Policies by Reconstructing Visual Tokens3 Feb
- 2025BadVLA: Towards Backdoor Attacks on Vision-Language-Action Models via Objective-Decoupled Optimization22 May
- 2024Exploring the Adversarial Vulnerabilities of Vision-Language-Action Models in Robotics18 Nov
Other references · 35
- 2017Attention Is All You Needcited by 60,361
- 2020An Image is Worth 16x16 Words: Transformers for Image Recognition at Scalecited by 30,186
- 2023Visual Instruction Tuningcited by 7,869
- 2023RT-2: Vision-Language-Action Models Transfer Web Knowledge to Robotic Controlcited by 3,080
- 2024π0: A Vision-Language-Action Flow Model for General Robot Controlcited by 2,407
- 2016Membership Inference Attacks against Machine Learning Modelscited by 2,281
- 2023LIBERO: Benchmarking Knowledge Transfer for Lifelong Robot Learningcited by 1,410
- 2025Fine-Tuning Vision-Language-Action Models: Optimizing Speed and Successcited by 883
- 2023Open X-Embodiment: Robotic Learning Datasets and RT-X Modelscited by 862
- 2025SpatialVLA: Exploring Spatial Representations for Visual-Language-Action Modelcited by 519
- 2025UniVLA: Learning to Act Anywhere with Task-centric Latent Actionscited by 410
- 2023Detecting Pretraining Data from Large Language Modelscited by 387
23 more
- 2022Deduplicating Training Data Mitigates Privacy Risks in Language Modelscited by 314
- 2025LIBERO-Plus: In-depth Robustness Analysis of Vision-Language-Action Modelscited by 303
- 2024A Survey on Vision-Language-Action Models for Embodied AIcited by 279
- 2025TWIST: Teleoperated Whole-Body Imitation Systemcited by 101
- 2025RoboCOIN: An Open-Sourced Bimanual Robotic Data Collection for Integrated Manipulationcited by 78
- 2025Towards Label-Only Membership Inference Attack against Pre-trained Large Language Modelscited by 37
- 2023Black-box Membership Inference Attacks against Fine-tuned Diffusion Modelscited by 35
- 2024Membership Inference Attacks against Large Vision-Language Modelscited by 33
- 2025Membership Inference Attacks Against Vision-Language Modelscited by 29
- 2026Psi0: An Open Foundation Model Towards Universal Humanoid Loco-Manipulationcited by 27
- 2025Exploring the limits of strong membership inference attacks on large language modelscited by 22
- 2025Membership Inference Attacks on Large-Scale Models: A Surveycited by 18
- 2025Membership Inference Attacks as Privacy Tools: Reliability, Disparity and Ensemblecited by 11
- 2024A Method to Facilitate Membership Inference Attacks in Deep Learning Modelscited by 7
- 2026VidLeaks: Membership Inference Attacks Against Text-to-Video Modelscited by 5
- 2025Black-Box Membership Inference Attack for LVLMs via Prior Knowledge-Calibrated Memory Probingcited by 4
- 2026AttenMIA: LLM Membership Inference Attack through Attention Signalscited by 4
- 2025Imitative Membership Inference Attackcited by 3
- 2026Toward Efficient Membership Inference Attacks against Federated Large Language Models: A Projection Residual Approachcited by 3
- 2026Inference Attacks Against Graph Generative Diffusion Modelscited by 2
- 2026When Reasoning Leaks Membership: Membership Inference Attack on Black-box Large Reasoning Modelscited by 2
- 2025Win-k: Improved Membership Inference Attacks on Small Language Modelscited by 1
- 2026A Kung Fu Athlete Bot That Can Do It All Day: Highly Dynamic, Balance-Challenging Motion Dataset and Autonomous Fall-Resilient Trackingcited by 1
Built on by
Citing papers per year: 2026: 2.
2 papers.
Similar, not cited
Close in wording, no citation either way
- 2025AttackVLA: Benchmarking Adversarial and Backdoor Attacks on Vision-Language-Action Models
- 2025Adversarial Attacks on Robotic Vision Language Action Models
- 2026Membership Inference Attacks on Vision-Language-Action Models
- 2026TrustVLA: Mechanism-Guided Inference-Time Defense Against Vision-Language-Action Backdoors
- 2025Phantom Menace: Exploring and Enhancing the Robustness of VLA Models Against Physical Sensor Attacks
- 2025OpenLVLM-MIA: A Controlled Benchmark Revealing the Limits of Membership Inference Attacks on Large Vision-Language Models
- 2025FreezeVLA: Action-Freezing Attacks against Vision-Language-Action Models
- 2026Towards Backdoor-Based Ownership Verification for Vision-Language-Action Models