AI Paper Atlas

Research map / Image generation / Text-to-image diffusion

Reason2Attack: Jailbreaking Text-to-Image Models via LLM Reasoning

Chenyu Zhang, Lanjun Wang, Yiwen Ma, Wenhui Li, Guoqing Jin, Anan Liu

AAAI 2026 · Image generation › Text-to-image diffusion · approach: T2I evaluation and controllable adapters

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 SFT on Frame Semantics CoT examples (ablation)Llama-3-8b-Instruct attacker LLM · SD1.4 + safety filtersaverage ASR: 0.49 -> 0.74Improved
add RL with attack result reward (ablation)Llama-3-8b-Instruct attacker LLM · SD1.4 + safety filtersaverage ASR: 0.49 -> 0.74Improved
use attack process reward instead of attack result reward in RL (ablation)RL-trained attacker LLM · SD1.4 + safety filtersaverage ASR: 0.74 -> 0.87Improved
use attack process reward instead of attack result reward in RL (ablation)RL-trained attacker LLM · SD1.4 + safety filtersaverage query number: 3.3 -> 2.6Improved
add SFT on CoT examples before RL with attack process reward (ablation)LLM+RL_AP · SD1.4 + safety filtersaverage ASR: 0.87 -> 0.90Improved
add SFT on CoT examples before RL with attack process reward (ablation)LLM+RL_AP · SD1.4 + safety filtersaverage query number: 2.6 -> 2.5Improved
add SFT on CoT examples before RL with attack process reward (ablation)LLM+RL_AP · SD1.4 + safety filtersaverage perplexity: 122 -> 155Hurt
use attack result reward only in RL (ablation)RL training of attacker LLM · RL return curvetraining stabilityHurt
R2A compared with iterative CMMA baselineR2A · SD1.4 + safety filtersaverage query number: 21.7 -> 2.5Improved
smaller gamma-style length/stealth trade-off factor (ablation)R2A · SD1.4 + safety filtersaverage ASRImproved
increase maximum number of queries (ablation)R2A · SD1.4 + safety filtersaverage ASRImproved
R2A on text-to-video model OpenSora V2 versus CMMAR2A · OpenSora V2average ASR: 0.59 -> 0.62Mixed

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