TAPDreamer: Transferable Adversarial Patches for World Action Models

Summary

TAPDreamer is an attack on world action models that uses only a public encoder to build a fixed local adversarial patch, requiring no target-policy queries. One frozen patch covering about 6.5% of the input cuts FastWAM's success from 97.7% to 0.0% on 40 LIBERO tasks and from 90.86% to 0.0% on 50 RoboTwin tasks, while matched random patches keep 81.5% and 79.2%. The authors conclude that defenses must also secure shared visual encoders against persistent local perturbations.

Topics

• Adversarial machine learning

Related items

• InfoA Unified Attack Framework against Statistical Generative Watermarks for Large Language ModelsSimilar attack · Arxiv (cs.CR + cs.CL + cs.LG)
• InfoExposing Vulnerability Monoculture Through Systemic Adversarial RobustnessSimilar attack · Arxiv (cs.CR + cs.CL + cs.LG)
• InfoTraining AIs to Solve CAPTCHAsSimilar attack · Schneier on Security
• InfoProvable Adversarial Detection: Prime Quantization meets Gromov--WassersteinSimilar attack · Arxiv (cs.CR + cs.CL + cs.LG)
• InfoAuditor-Assisted Summary-Channel Verification for Hosted LLM Identity SubstitutionSimilar attack · Arxiv (cs.CR + cs.CL + cs.LG)