# Audio Jailbreaking on HarmBench Standard category

Source: [Audio Jailbreaking on HarmBench Standard category](<https://www.sota2.com/research/sota/audio-jailbreaking-on-harmbench-standard-category>)

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

Data updated: 2026-05-07T08:58:11.616Z

Evaluation of model susceptibility to jailbreaking attacks specifically targeting audio inputs within the HarmBench Standard category, reporting audio-specific success rates (ASRr/ASRl).

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 | Audio Jailbreaking |
| Dataset (including any stated split) | HarmBench Standard category |
| Metric | ASRr |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 9 of 9 |
| 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

- [ASRr](<https://www.sota2.com/research/sota/audio-jailbreaking-on-harmbench-standard-category.md?metric=ASRr>): Higher is better; unit: PERCENT.

- [ASRl](<https://www.sota2.com/research/sota/audio-jailbreaking-on-harmbench-standard-category.md?metric=ASRl>): Higher is better; unit: PERCENT.

## Recorded results

| Method | ASRr | Result date | Report paper |
| --- | --- | --- | --- |
| TAGO | 100 | Not recorded | 69fc1cb75497b3b670c886e2 |
| TAGO | 100 | Not recorded | 69fc1cb75497b3b670c886e2 |
| TAGO | 100 | Not recorded | 69fc1cb75497b3b670c886e2 |
| TAGO | 100 | Not recorded | 69fc1cb75497b3b670c886e2 |
| TAGO | 99 | Not recorded | 69fc1cb75497b3b670c886e2 |
| TAGO | 98 | Not recorded | 69fc1cb75497b3b670c886e2 |
| Direct | 97 | Not recorded | 69fc1cb75497b3b670c886e2 |
| Direct | 40.5 | Not recorded | 69fc1cb75497b3b670c886e2 |
| Direct | 5.5 | Not recorded | 69fc1cb75497b3b670c886e2 |

### TAGO

ASRr: 100. Result ID: 69fc541547f34ce924df2f83.

[Reporting paper](<https://www.sota2.com/research/paper/69fc1cb75497b3b670c886e2.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69fc1cb75497b3b670c886e2.md>)

Recorded evaluation configuration: {"Model":"Qwen3-Omni","Token retention ratio (ζ)":"1.0"}. This record does not assert independent reproduction.

### TAGO

ASRr: 100. Result ID: 69fc541547f34ce924df2f86.

[Reporting paper](<https://www.sota2.com/research/paper/69fc1cb75497b3b670c886e2.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69fc1cb75497b3b670c886e2.md>)

Recorded evaluation configuration: {"Model":"Qwen2.5-Omni","Token retention ratio (ζ)":"1.0"}. This record does not assert independent reproduction.

### TAGO

ASRr: 100. Result ID: 69fc541547f34ce924df2f89.

[Reporting paper](<https://www.sota2.com/research/paper/69fc1cb75497b3b670c886e2.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69fc1cb75497b3b670c886e2.md>)

Recorded evaluation configuration: {"Model":"LLaMA-Omni","Token retention ratio (ζ)":"1.0"}. This record does not assert independent reproduction.

### TAGO

ASRr: 100. Result ID: 69fc541547f34ce924df2f8a.

[Reporting paper](<https://www.sota2.com/research/paper/69fc1cb75497b3b670c886e2.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69fc1cb75497b3b670c886e2.md>)

Recorded evaluation configuration: {"Model":"LLaMA-Omni","Token retention ratio (ζ)":"0.25"}. This record does not assert independent reproduction.

### TAGO

ASRr: 99. Result ID: 69fc541547f34ce924df2f84.

[Reporting paper](<https://www.sota2.com/research/paper/69fc1cb75497b3b670c886e2.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69fc1cb75497b3b670c886e2.md>)

Recorded evaluation configuration: {"Model":"Qwen3-Omni","Token retention ratio (ζ)":"0.25"}. This record does not assert independent reproduction.

### TAGO

ASRr: 98. Result ID: 69fc541547f34ce924df2f87.

[Reporting paper](<https://www.sota2.com/research/paper/69fc1cb75497b3b670c886e2.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69fc1cb75497b3b670c886e2.md>)

Recorded evaluation configuration: {"Model":"Qwen2.5-Omni","Token retention ratio (ζ)":"0.25"}. This record does not assert independent reproduction.

### Direct

ASRr: 97. Result ID: 69fc541547f34ce924df2f88.

[Reporting paper](<https://www.sota2.com/research/paper/69fc1cb75497b3b670c886e2.md>)

Recorded evaluation configuration: {"Model":"LLaMA-Omni"}. This record does not assert independent reproduction.

### Direct

ASRr: 40.5. Result ID: 69fc541547f34ce924df2f85.

[Reporting paper](<https://www.sota2.com/research/paper/69fc1cb75497b3b670c886e2.md>)

Recorded evaluation configuration: {"Model":"Qwen2.5-Omni"}. This record does not assert independent reproduction.

### Direct

ASRr: 5.5. Result ID: 69fc541547f34ce924df2f82.

[Reporting paper](<https://www.sota2.com/research/paper/69fc1cb75497b3b670c886e2.md>)

Recorded evaluation configuration: {"Model":"Qwen3-Omni"}. This record does not assert independent reproduction.

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