Universal Cross-Prompt Adversarial Attacks on Promptable Concept Segmentation

Summary

Researchers present AdvPCS, a universal cross-prompt adversarial attack on Promptable Concept Segmentation in SAM3, which extends SAM-series models to concept-level prediction. The method combines min-max prompt optimization with perception deception and temporal memory misalignment attacks. A single universal adversarial perturbation (UAP) generalizes across frames from different videos and reduces the average mIoU of PCS models on the SA-CO dataset to below 5% under text prompts.

Topics

• Adversarial machine learning

Related items

• InfoCan you trust Meta’s Muse or OpenAI’s Dots to run your life?Same vendor · The Verge (AI)
• InfoAI startup Manus raises $500 million in first funding round since Meta breakupSame vendor · CNBC Technology
• InfoMicrosoft to sell $2,599 Surface Laptop Ultra containing Nvidia AI chipSame vendor · CNBC Technology
• InfoIt appears .agent and .agi are about to be the hot new domainsSame vendor · The Verge (AI)
• InfoMuse launches on the iPadSame vendor · The Verge (AI)