# Text-to-Image Jailbreak Attack on NSFW-200

Source: [Text-to-Image Jailbreak Attack on NSFW-200](<https://www.sota2.com/research/sota/text-to-image-jailbreak-attack-on-nsfw-200>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-06-24T07:40:57.687Z

Evaluation of model resistance to jailbreak attacks targeting text-to-image generation capabilities using the NSFW-200 dataset, measured by success rates in bypassing safety mechanisms.

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

| Scope | Recorded value |
| --- | --- |
| Task | Text-to-Image Jailbreak Attack |
| Dataset (including any stated split) | NSFW-200 |
| Metric | Attack Success Rate (Paper) |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 3 of 3 |
| Coverage | All results currently returned for this benchmark |

Evaluation settings may differ between results. The original papers and recorded configurations are needed to establish comparability. A shared metric name alone does not establish an identical protocol.

## Metrics

- [Attack Success Rate (Paper)](<https://www.sota2.com/research/sota/text-to-image-jailbreak-attack-on-nsfw-200.md?metric=Attack%20Success%20Rate%20(Paper)>): Higher is better; unit: PERCENT.

- [Attack Success Rate (Gen)](<https://www.sota2.com/research/sota/text-to-image-jailbreak-attack-on-nsfw-200.md?metric=Attack%20Success%20Rate%20(Gen)>): Higher is better; unit: PERCENT.

- [Difference in Attack Success Rate (Paper - Gen)](<https://www.sota2.com/research/sota/text-to-image-jailbreak-attack-on-nsfw-200.md?metric=Difference%20in%20Attack%20Success%20Rate%20(Paper%20-%20Gen)>): Lower is better; unit: PERCENT.

## Recorded results

| Method | Attack Success Rate (Paper) | Result date | Report paper |
| --- | --- | --- | --- |
| SneakyPrompt | 100 | Not recorded | 6a3b647c206542251d7c9622 |
| JailFuzzer | 100 | Not recorded | 6a3b647c206542251d7c9622 |
| PGJ | 100 | Not recorded | 6a3b647c206542251d7c9622 |

### SneakyPrompt

Attack Success Rate (Paper): 100. Result ID: 6a3b89eefbc373592c43cb86.

[Reporting paper](<https://www.sota2.com/research/paper/6a3b647c206542251d7c9622.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa712f1e209131cea4fd09.md>)

Recorded evaluation configuration: {"Evaluation Metric":"Bypass Rate","Code Ref":"YES","Victim Model":"SDv1.4","ρ (LoC difference)":"-1075"}. This record does not assert independent reproduction.

### JailFuzzer

Attack Success Rate (Paper): 100. Result ID: 6a3b89eefbc373592c43cb87.

[Reporting paper](<https://www.sota2.com/research/paper/6a3b647c206542251d7c9622.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa79ac1e209131cea7ed1f.md>)

Recorded evaluation configuration: {"Evaluation Metric":"Bypass Rate","Code Ref":"YES","Victim Model":"SDv1.4","ρ (LoC difference)":"-1309"}. This record does not assert independent reproduction.

### PGJ

Attack Success Rate (Paper): 100. Result ID: 6a3b89eefbc373592c43cb88.

[Reporting paper](<https://www.sota2.com/research/paper/6a3b647c206542251d7c9622.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa76e01e209131cea6f86e.md>)

Recorded evaluation configuration: {"Evaluation Metric":"ASR-1","Code Ref":"NO","Victim Model":"SDXL","ρ (LoC difference)":"-"}. This record does not assert independent reproduction.

Page 1 of 1. This is an index page, not the full collection.