AI Paper Atlas

Research map / Adversarial robustness and security / Attacks on text and graphs

ARMs: Adaptive Red-Teaming Agent against Multimodal Models with Plug-and-Play Attacks

Zhaorun Chen, Xun Liu, Mintong Kang, Jiawei Zhang, Minzhou Pan, Shuang Yang, Bo Li

ICLR 2026 · Adversarial robustness and security › Attacks on text and graphs · approach: Vision-language model robustness · 6 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add layered memory retrieval (top-k=3) to ARMs (ablation)ARMs · StrongReject vs Claude-3.7-SonnetASR: 89.4 -> 95.2Improved
retrieve 7 memories instead of 3 (ablation)ARMs · StrongReject vs Claude-3.7-SonnetASR: 95.2 -> 85.6Hurt
add epsilon-greedy decay (lambda=1) instead of pure exploration (ablation)ARMs · StrongReject vs Claude-3.7-SonnetASR: 86.0 -> 95.2Improved
add vision modality input to the backbone (ablation)ARMs · StrongReject vs Claude-3.7-SonnetASR: 92.1 -> 95.2Improved
add reasoning instructions to the backbone (ablation)ARMs · StrongReject vs Claude-3.7-SonnetASR: 82.9 -> 95.2Improved
use GPT-4o instead of Qwen3-235B as backbone (ablation)ARMs · StrongReject vs Claude-3.7-SonnetASR: 80.6 -> 95.2Improved
use ARMs multi-step orchestration instead of brute-force oracle over all strategiesARMs · StrongReject vs Claude-3.7-SonnetASR: 84.0 -> 95.2Improved
use o3-mini judge instead of GPT-4o judge (ablation)ARMs · EU AI Act vs GPT-4o victimASR: 76.9 -> 100.0Improved
use Claude-3.7-Sonnet backbone instead of GPT-4o (ablation)ARMs · StrongReject vs GPT-4o victimASR: 93.1 -> 99.4Improved
fine-tune InternVL3-38B with ARMs-Bench safety dataInternVL3-38B · StrongReject ARMs ASRASR: 98.5 -> 69.6Improved
fine-tune InternVL3-38B with ARMs-Bench safety dataInternVL3-38B · Policy-based ARMs ASRASR: 87.7 -> 29.2Improved
fine-tune InternVL3-38B with JailbreakV safety dataInternVL3-38B · StrongReject ARMs ASRASR: 98.5 -> 98.0No clear effect

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy