# Multimodal Jailbreak Attack on Image-generation prompt-sets

Source: [Multimodal Jailbreak Attack on Image-generation prompt-sets](<https://www.sota2.com/research/sota/multimodal-jailbreak-attack-on-image-generation-prompt-sets>)

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

Data updated: 2026-04-08T07:07:26.052Z

Evaluation of model resistance to multimodal jailbreak attacks using the Image-generation prompt-sets, measured by AASR and AARR.

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 | Multimodal Jailbreak Attack |
| Dataset (including any stated split) | Image-generation prompt-sets |
| Metric | AASR |
| Unit | PERCENT |
| Direction | Lower is better |
| Results loaded | 18 of 18 |
| 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

- [AASR](<https://www.sota2.com/research/sota/multimodal-jailbreak-attack-on-image-generation-prompt-sets.md?metric=AASR>): Lower is better; unit: PERCENT.

- [AARR](<https://www.sota2.com/research/sota/multimodal-jailbreak-attack-on-image-generation-prompt-sets.md?metric=AARR>): Lower is better; unit: PERCENT.

## Recorded results

| Method | AASR | Result date | Report paper |
| --- | --- | --- | --- |
| Gemini-2.5 | 19 | Not recorded | 68fa7c0e1e209131cea8c27a |
| ChatGPT | 20 | Not recorded | 68fa7c0e1e209131cea8c27a |
| Gemini-2.5 | 24 | Not recorded | 68fa7c0e1e209131cea8c27a |
| ChatGPT | 29 | Not recorded | 68fa7c0e1e209131cea8c27a |
| Gemini-2.5 | 30 | Not recorded | 68fa7c0e1e209131cea8c27a |
| ChatGPT | 35 | Not recorded | 68fa7c0e1e209131cea8c27a |
| Gemini-2.5 | 38 | Not recorded | 68fa7c0e1e209131cea8c27a |
| Gemini-2.5 | 40 | Not recorded | 68fa7c0e1e209131cea8c27a |
| Gemini-2.5 | 43 | Not recorded | 68fa7c0e1e209131cea8c27a |
| ChatGPT | 45 | Not recorded | 68fa7c0e1e209131cea8c27a |
| Nano Banana Pro | 45 | Not recorded | 68fa7c0e1e209131cea8c27a |
| Nano Banana Pro | 48 | Not recorded | 68fa7c0e1e209131cea8c27a |
| Nano Banana Pro | 63 | Not recorded | 68fa7c0e1e209131cea8c27a |
| ChatGPT | 65 | Not recorded | 68fa7c0e1e209131cea8c27a |
| Nano Banana Pro | 65 | Not recorded | 68fa7c0e1e209131cea8c27a |
| Nano Banana Pro | 69 | Not recorded | 68fa7c0e1e209131cea8c27a |
| Nano Banana Pro | 76 | Not recorded | 68fa7c0e1e209131cea8c27a |
| ChatGPT | 78 | Not recorded | 68fa7c0e1e209131cea8c27a |

### Gemini-2.5

AASR: 19. Result ID: 69d5fea547f34ce924be1675.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7c0e1e209131cea8c27a.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7c311e209131cea8cef5.md>)

Recorded evaluation configuration: {"Jailbreak Template":"Base","Input Set":"CM"}. This record does not assert independent reproduction.

### ChatGPT

AASR: 20. Result ID: 69d5fea547f34ce924be166e.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7c0e1e209131cea8c27a.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa75681e209131cea67a98.md>)

Recorded evaluation configuration: {"Jailbreak Template":"Base","Input Set":"Eng"}. This record does not assert independent reproduction.

### Gemini-2.5

AASR: 24. Result ID: 69d5fea547f34ce924be1676.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7c0e1e209131cea8c27a.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7c311e209131cea8cef5.md>)

Recorded evaluation configuration: {"Jailbreak Template":"Base","Input Set":"CMP"}. This record does not assert independent reproduction.

### ChatGPT

AASR: 29. Result ID: 69d5fea547f34ce924be166f.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7c0e1e209131cea8c27a.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa75681e209131cea67a98.md>)

Recorded evaluation configuration: {"Jailbreak Template":"Base","Input Set":"CM"}. This record does not assert independent reproduction.

### Gemini-2.5

AASR: 30. Result ID: 69d5fea547f34ce924be1674.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7c0e1e209131cea8c27a.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7c311e209131cea8cef5.md>)

Recorded evaluation configuration: {"Jailbreak Template":"Base","Input Set":"Eng"}. This record does not assert independent reproduction.

### ChatGPT

AASR: 35. Result ID: 69d5fea547f34ce924be1671.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7c0e1e209131cea8c27a.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa75681e209131cea67a98.md>)

Recorded evaluation configuration: {"Jailbreak Template":"VisLM","Input Set":"Eng"}. This record does not assert independent reproduction.

### Gemini-2.5

AASR: 38. Result ID: 69d5fea547f34ce924be1677.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7c0e1e209131cea8c27a.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7c311e209131cea8cef5.md>)

Recorded evaluation configuration: {"Jailbreak Template":"VisLM","Input Set":"Eng"}. This record does not assert independent reproduction.

### Gemini-2.5

AASR: 40. Result ID: 69d5fea547f34ce924be1678.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7c0e1e209131cea8c27a.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7c311e209131cea8cef5.md>)

Recorded evaluation configuration: {"Jailbreak Template":"VisLM","Input Set":"CM"}. This record does not assert independent reproduction.

### Gemini-2.5

AASR: 43. Result ID: 69d5fea547f34ce924be1679.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7c0e1e209131cea8c27a.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7c311e209131cea8cef5.md>)

Recorded evaluation configuration: {"Jailbreak Template":"VisLM","Input Set":"CMP"}. This record does not assert independent reproduction.

### ChatGPT

AASR: 45. Result ID: 69d5fea547f34ce924be1672.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7c0e1e209131cea8c27a.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa75681e209131cea67a98.md>)

Recorded evaluation configuration: {"Jailbreak Template":"VisLM","Input Set":"CM"}. This record does not assert independent reproduction.

### Nano Banana Pro

AASR: 45. Result ID: 69d5fea547f34ce924be167a.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7c0e1e209131cea8c27a.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7c311e209131cea8cef5.md>)

Recorded evaluation configuration: {"Jailbreak Template":"Base","Input Set":"Eng"}. This record does not assert independent reproduction.

### Nano Banana Pro

AASR: 48. Result ID: 69d5fea547f34ce924be167b.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7c0e1e209131cea8c27a.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7c311e209131cea8cef5.md>)

Recorded evaluation configuration: {"Jailbreak Template":"Base","Input Set":"CM"}. This record does not assert independent reproduction.

### Nano Banana Pro

AASR: 63. Result ID: 69d5fea547f34ce924be167d.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7c0e1e209131cea8c27a.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7c311e209131cea8cef5.md>)

Recorded evaluation configuration: {"Jailbreak Template":"VisLM","Input Set":"Eng"}. This record does not assert independent reproduction.

### ChatGPT

AASR: 65. Result ID: 69d5fea547f34ce924be1670.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7c0e1e209131cea8c27a.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa75681e209131cea67a98.md>)

Recorded evaluation configuration: {"Jailbreak Template":"Base","Input Set":"CMP"}. This record does not assert independent reproduction.

### Nano Banana Pro

AASR: 65. Result ID: 69d5fea547f34ce924be167c.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7c0e1e209131cea8c27a.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7c311e209131cea8cef5.md>)

Recorded evaluation configuration: {"Jailbreak Template":"Base","Input Set":"CMP"}. This record does not assert independent reproduction.

### Nano Banana Pro

AASR: 69. Result ID: 69d5fea547f34ce924be167e.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7c0e1e209131cea8c27a.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7c311e209131cea8cef5.md>)

Recorded evaluation configuration: {"Jailbreak Template":"VisLM","Input Set":"CM"}. This record does not assert independent reproduction.

### Nano Banana Pro

AASR: 76. Result ID: 69d5fea547f34ce924be167f.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7c0e1e209131cea8c27a.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7c311e209131cea8cef5.md>)

Recorded evaluation configuration: {"Jailbreak Template":"VisLM","Input Set":"CMP"}. This record does not assert independent reproduction.

### ChatGPT

AASR: 78. Result ID: 69d5fea547f34ce924be1673.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7c0e1e209131cea8c27a.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa75681e209131cea67a98.md>)

Recorded evaluation configuration: {"Jailbreak Template":"VisLM","Input Set":"CMP"}. This record does not assert independent reproduction.

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