# Jailbreak Defense on AdvBench (GCG Attack)

Source: [Jailbreak Defense on AdvBench (GCG Attack)](<https://www.sota2.com/research/sota/jailbreak-defense-on-advbench-gcg-attack>)

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

Data updated: 2026-02-26T05:40:31.933Z

Evaluation of model performance defending against jailbreak prompts generated using the GCG attack methodology on the AdvBench dataset, measured by Defense Success Rate (DSR).

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 Defense |
| Dataset (including any stated split) | AdvBench GCG attack |
| Metric | DSR |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 15 of 15 |
| 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

- [DSR](<https://www.sota2.com/research/sota/jailbreak-defense-on-advbench-gcg-attack.md?metric=DSR>): Higher is better; unit: PERCENT.

## Recorded results

| Method | DSR | Result date | Report paper |
| --- | --- | --- | --- |
| SmoothLLM | 100 | Not recorded | 68fa738f1e209131cea5d3a7 |
| Paraphrasing | 100 | Not recorded | 68fa738f1e209131cea5d3a7 |
| Backtranslation | 100 | Not recorded | 68fa738f1e209131cea5d3a7 |
| Response check | 100 | Not recorded | 68fa738f1e209131cea5d3a7 |
| Backtranslation | 100 | Not recorded | 68fa738f1e209131cea5d3a7 |
| SmoothLLM | 98 | Not recorded | 68fa738f1e209131cea5d3a7 |
| Paraphrasing | 98 | Not recorded | 68fa738f1e209131cea5d3a7 |
| Backtranslation | 98 | Not recorded | 68fa738f1e209131cea5d3a7 |
| No defense | 94 | Not recorded | 68fa738f1e209131cea5d3a7 |
| Response check | 94 | Not recorded | 68fa738f1e209131cea5d3a7 |
| SmoothLLM | 92 | Not recorded | 68fa738f1e209131cea5d3a7 |
| Paraphrasing | 84 | Not recorded | 68fa738f1e209131cea5d3a7 |
| No defense | 66 | Not recorded | 68fa738f1e209131cea5d3a7 |
| Response check | 30 | Not recorded | 68fa738f1e209131cea5d3a7 |
| No defense | 8 | Not recorded | 68fa738f1e209131cea5d3a7 |

### SmoothLLM

DSR: 100. Result ID: 699300653a6c31484912f775.

[Reporting paper](<https://www.sota2.com/research/paper/68fa738f1e209131cea5d3a7.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa73961e209131cea5d658.md>)

Recorded evaluation configuration: {"Target Model":"GPT-3.5-turbo"}. This record does not assert independent reproduction.

### Paraphrasing

DSR: 100. Result ID: 699300653a6c31484912f776.

[Reporting paper](<https://www.sota2.com/research/paper/68fa738f1e209131cea5d3a7.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa707f1e209131cea4b7c6.md>)

Recorded evaluation configuration: {"Target Model":"GPT-3.5-turbo"}. This record does not assert independent reproduction.

### Backtranslation

DSR: 100. Result ID: 699300653a6c31484912f778.

[Reporting paper](<https://www.sota2.com/research/paper/68fa738f1e209131cea5d3a7.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa738f1e209131cea5d3a7.md>)

[Implementation](<https://github.com/YihanWang617/LLM-Jailbreaking-Defense-Backtranslation>)

Recorded evaluation configuration: {"Target Model":"GPT-3.5-turbo"}. This record does not assert independent reproduction.

### Response check

DSR: 100. Result ID: 699300653a6c31484912f77c.

[Reporting paper](<https://www.sota2.com/research/paper/68fa738f1e209131cea5d3a7.md>)

Recorded evaluation configuration: {"Target Model":"Llama-2-13B"}. This record does not assert independent reproduction.

### Backtranslation

DSR: 100. Result ID: 699300653a6c31484912f77d.

[Reporting paper](<https://www.sota2.com/research/paper/68fa738f1e209131cea5d3a7.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa738f1e209131cea5d3a7.md>)

[Implementation](<https://github.com/YihanWang617/LLM-Jailbreaking-Defense-Backtranslation>)

Recorded evaluation configuration: {"Target Model":"Llama-2-13B"}. This record does not assert independent reproduction.

### SmoothLLM

DSR: 98. Result ID: 699300653a6c31484912f77a.

[Reporting paper](<https://www.sota2.com/research/paper/68fa738f1e209131cea5d3a7.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa73961e209131cea5d658.md>)

Recorded evaluation configuration: {"Target Model":"Llama-2-13B"}. This record does not assert independent reproduction.

### Paraphrasing

DSR: 98. Result ID: 699300653a6c31484912f77b.

[Reporting paper](<https://www.sota2.com/research/paper/68fa738f1e209131cea5d3a7.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa707f1e209131cea4b7c6.md>)

Recorded evaluation configuration: {"Target Model":"Llama-2-13B"}. This record does not assert independent reproduction.

### Backtranslation

DSR: 98. Result ID: 699300653a6c31484912f782.

[Reporting paper](<https://www.sota2.com/research/paper/68fa738f1e209131cea5d3a7.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa738f1e209131cea5d3a7.md>)

[Implementation](<https://github.com/YihanWang617/LLM-Jailbreaking-Defense-Backtranslation>)

Recorded evaluation configuration: {"Target Model":"Vicuna-13B"}. This record does not assert independent reproduction.

### No defense

DSR: 94. Result ID: 699300653a6c31484912f774.

[Reporting paper](<https://www.sota2.com/research/paper/68fa738f1e209131cea5d3a7.md>)

Recorded evaluation configuration: {"Target Model":"GPT-3.5-turbo"}. This record does not assert independent reproduction.

### Response check

DSR: 94. Result ID: 699300653a6c31484912f777.

[Reporting paper](<https://www.sota2.com/research/paper/68fa738f1e209131cea5d3a7.md>)

Recorded evaluation configuration: {"Target Model":"GPT-3.5-turbo"}. This record does not assert independent reproduction.

### SmoothLLM

DSR: 92. Result ID: 699300653a6c31484912f77f.

[Reporting paper](<https://www.sota2.com/research/paper/68fa738f1e209131cea5d3a7.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa73961e209131cea5d658.md>)

Recorded evaluation configuration: {"Target Model":"Vicuna-13B"}. This record does not assert independent reproduction.

### Paraphrasing

DSR: 84. Result ID: 699300653a6c31484912f780.

[Reporting paper](<https://www.sota2.com/research/paper/68fa738f1e209131cea5d3a7.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa707f1e209131cea4b7c6.md>)

Recorded evaluation configuration: {"Target Model":"Vicuna-13B"}. This record does not assert independent reproduction.

### No defense

DSR: 66. Result ID: 699300653a6c31484912f779.

[Reporting paper](<https://www.sota2.com/research/paper/68fa738f1e209131cea5d3a7.md>)

Recorded evaluation configuration: {"Target Model":"Llama-2-13B"}. This record does not assert independent reproduction.

### Response check

DSR: 30. Result ID: 699300653a6c31484912f781.

[Reporting paper](<https://www.sota2.com/research/paper/68fa738f1e209131cea5d3a7.md>)

Recorded evaluation configuration: {"Target Model":"Vicuna-13B"}. This record does not assert independent reproduction.

### No defense

DSR: 8. Result ID: 699300653a6c31484912f77e.

[Reporting paper](<https://www.sota2.com/research/paper/68fa738f1e209131cea5d3a7.md>)

Recorded evaluation configuration: {"Target Model":"Vicuna-13B"}. This record does not assert independent reproduction.

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