# Jailbreak Attack on SPA-VL Harm

Source: [Jailbreak Attack on SPA-VL Harm](<https://www.sota2.com/research/sota/jailbreak-attack-on-spa-vl-harm>)

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

Data updated: 2026-02-26T05:23:56.771Z

Evaluation of model performance against jailbreak attacks using the SPA-VL Harm dataset, measured by Attack Success Rate (ASR).

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 | Jailbreak Attack |
| Dataset (including any stated split) | SPA-VL Harm |
| Metric | ASR |
| Unit | PERCENT |
| Direction | Lower is better |
| Results loaded | 5 of 5 |
| 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

- [ASR](<https://www.sota2.com/research/sota/jailbreak-attack-on-spa-vl-harm.md?metric=ASR>): Lower is better; unit: PERCENT.

## Recorded results

| Method | ASR | Result date | Report paper |
| --- | --- | --- | --- |
| Original | 10.57 | Not recorded | 69802fbba8b01e76cc8e1718 |
| Only Image Noising | 13.96 | Not recorded | 69802fbba8b01e76cc8e1718 |
| Static Rewriting | 24.15 | Not recorded | 69802fbba8b01e76cc8e1718 |
| ReAct Dynamic Prompt Rewriting | 49.81 | Not recorded | 69802fbba8b01e76cc8e1718 |
| ReAct-driven adaptive noising mechanism | 52.08 | Not recorded | 69802fbba8b01e76cc8e1718 |

### Original

ASR: 10.57. Result ID: 6997762b5698665ec4ca58e1.

[Reporting paper](<https://www.sota2.com/research/paper/69802fbba8b01e76cc8e1718.md>)

Recorded evaluation configuration: {"mode":"baseline"}. This record does not assert independent reproduction.

### Only Image Noising

ASR: 13.96. Result ID: 6997762b5698665ec4ca58e3.

[Reporting paper](<https://www.sota2.com/research/paper/69802fbba8b01e76cc8e1718.md>)

Recorded evaluation configuration: {"strategy":"Adversarial image noising"}. This record does not assert independent reproduction.

### Static Rewriting

ASR: 24.15. Result ID: 6997762b5698665ec4ca58e2.

[Reporting paper](<https://www.sota2.com/research/paper/69802fbba8b01e76cc8e1718.md>)

Recorded evaluation configuration: {"strategy":"Static template rewriting"}. This record does not assert independent reproduction.

### ReAct Dynamic Prompt Rewriting

ASR: 49.81. Result ID: 6997762b5698665ec4ca58e4.

[Reporting paper](<https://www.sota2.com/research/paper/69802fbba8b01e76cc8e1718.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69802fbba8b01e76cc8e1718.md>)

Recorded evaluation configuration: {"mode":"Only ReAct Rewriting"}. This record does not assert independent reproduction.

### ReAct-driven adaptive noising mechanism

ASR: 52.08. Result ID: 6997762b5698665ec4ca58e5.

[Reporting paper](<https://www.sota2.com/research/paper/69802fbba8b01e76cc8e1718.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69802fbba8b01e76cc8e1718.md>)

Recorded evaluation configuration: {"mode":"ReAct & Noise"}. This record does not assert independent reproduction.

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