cs.CVNov 15, 2026

When Backdoors Can Be Switched Off: Suppression-Aware Backdoor Attacks on Vision-Language-Action Models

Authors: Nur Afsa Syeda, Mohamed Elmahallawy

Organizations: School of Engineering and Applied Sciences, Washington State University, Richland, WA 99354, USA

Abstract

Vision-Language-Action (VLA) models map visual observations, robot states, and natural-language instructions to robotic actions, but their multimodal interfaces introduce new security risks. Existing VLA backdoor attacks primarily study trigger-based activation and stealthy action manipulation, leaving conditional suppression of malicious behavior underexplored. We investigate suppression-aware backdoors, which combine conjunctive trigger activation with instruction-conditioned suppression. We instantiate this design through three variants: Cross-Modal Trigger, combining robot-state and spatially separated visual triggers; Self-Suppressing Trigger, requiring a state trigger and a [DISRUPT] instruction token for activation, with [RESTORE] token suppressing the attack; and Multi-Trigger, additionally requiring a visual trigger while retaining suppression even when physical triggers persist. We evaluate these attacks using SmolVLA on two SO-100 manipulation datasets against four baselines: State Backdoor, BadVLA, TabVLA, and SilentDrift. Self-Suppressing and Multi-Trigger reduce Isolation Forest detection from 96--98% to 1--18% and action-space l divergence by 10--14X, achieving 63--71% attack success rates with perfect clean-task performance. Experiments with OpenVLA on LIBERO further show that the attacks transfer to a 7B-parameter policy, evading STRIP and action-distribution-shift detection. Our findings establish conditional suppression as an important dimension for evaluating backdoor threats in VLA policies.

Explore similar work

CardsList
  1. DropVLA: An Action-Level Backdoor Attack on Vision-Language-Action Models

    Oct 13, 2025Zonghuan Xu, Jiayu Li, Yunhan Zhao +3Adversarial Attacks on VLA ModelsRobotic Manipulation

  2. TrustVLA: Mechanism-Guided Inference-Time Defense Against Vision-Language-Action Backdoors

    Jul 14, 2026Pinhan Fu, Xianda Guo, Xuetao Li +5Adversarial Attacks on VLA ModelsVision-Language-Action Models

  3. Towards Backdoor-Based Ownership Verification for Vision-Language-Action Models

    May 9, 2026Ming Sun, Rui Wang, Xingrui Yu +5Vision-Language-Action ModelsGNN Watermarking