# Safety Evaluation (Attack Success Rate) on Pixverse

Source: [Safety Evaluation (Attack Success Rate) on Pixverse](<https://www.sota2.com/research/sota/safety-evaluation-attack-success-rate-on-pixverse>)

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

Data updated: 2026-03-10T08:31:00.035Z

Evaluation of model safety performance across various harmful content categories (e.g., Pornography, Violence, Discrimination) using the Pixverse dataset.

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 | Safety Evaluation (Attack Success Rate) |
| Dataset (including any stated split) | Pixverse |
| Metric | Pornography 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

- [Pornography ASR](<https://www.sota2.com/research/sota/safety-evaluation-attack-success-rate-on-pixverse.md?metric=Pornography%20ASR>): Lower is better; unit: PERCENT.

- [Borderline Pornography ASR](<https://www.sota2.com/research/sota/safety-evaluation-attack-success-rate-on-pixverse.md?metric=Borderline%20Pornography%20ASR>): Lower is better; unit: PERCENT.

- [Violence ASR](<https://www.sota2.com/research/sota/safety-evaluation-attack-success-rate-on-pixverse.md?metric=Violence%20ASR>): Lower is better; unit: PERCENT.

- [Gore ASR](<https://www.sota2.com/research/sota/safety-evaluation-attack-success-rate-on-pixverse.md?metric=Gore%20ASR>): Lower is better; unit: PERCENT.

- [Disturbing Content ASR](<https://www.sota2.com/research/sota/safety-evaluation-attack-success-rate-on-pixverse.md?metric=Disturbing%20Content%20ASR>): Lower is better; unit: PERCENT.

- [Public Figures ASR](<https://www.sota2.com/research/sota/safety-evaluation-attack-success-rate-on-pixverse.md?metric=Public%20Figures%20ASR>): Lower is better; unit: PERCENT.

- [Discrimination ASR](<https://www.sota2.com/research/sota/safety-evaluation-attack-success-rate-on-pixverse.md?metric=Discrimination%20ASR>): Lower is better; unit: PERCENT.

- [Political Sensitivity ASR](<https://www.sota2.com/research/sota/safety-evaluation-attack-success-rate-on-pixverse.md?metric=Political%20Sensitivity%20ASR>): Lower is better; unit: PERCENT.

- [Copyright ASR](<https://www.sota2.com/research/sota/safety-evaluation-attack-success-rate-on-pixverse.md?metric=Copyright%20ASR>): Lower is better; unit: PERCENT.

- [Illegal Activities ASR](<https://www.sota2.com/research/sota/safety-evaluation-attack-success-rate-on-pixverse.md?metric=Illegal%20Activities%20ASR>): Lower is better; unit: PERCENT.

- [Misinformation ASR](<https://www.sota2.com/research/sota/safety-evaluation-attack-success-rate-on-pixverse.md?metric=Misinformation%20ASR>): Lower is better; unit: PERCENT.

- [Sequential Action ASR](<https://www.sota2.com/research/sota/safety-evaluation-attack-success-rate-on-pixverse.md?metric=Sequential%20Action%20ASR>): Lower is better; unit: PERCENT.

- [Dynamic Variation ASR](<https://www.sota2.com/research/sota/safety-evaluation-attack-success-rate-on-pixverse.md?metric=Dynamic%20Variation%20ASR>): Lower is better; unit: PERCENT.

- [Coherent Contextual ASR](<https://www.sota2.com/research/sota/safety-evaluation-attack-success-rate-on-pixverse.md?metric=Coherent%20Contextual%20ASR>): Lower is better; unit: PERCENT.

- [Average ASR](<https://www.sota2.com/research/sota/safety-evaluation-attack-success-rate-on-pixverse.md?metric=Average%20ASR>): Lower is better; unit: PERCENT.

## Recorded results

| Method | Pornography ASR | Result date | Report paper |
| --- | --- | --- | --- |
| TSB | 14 | Not recorded | 69afa5662d603e6de3e43878 |
| RAB | 28 | Not recorded | 69afa5662d603e6de3e43878 |
| DACA | 28 | Not recorded | 69afa5662d603e6de3e43878 |
| VEIL | 80 | Not recorded | 69afa5662d603e6de3e43878 |
| TFM | 90 | Not recorded | 69afa5662d603e6de3e43878 |

### TSB

Pornography ASR: 14. Result ID: 69afd6c06dbb7075b958300a.

[Reporting paper](<https://www.sota2.com/research/paper/69afa5662d603e6de3e43878.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa74bb1e209131cea63572.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### RAB

Pornography ASR: 28. Result ID: 69afd6c06dbb7075b958300b.

[Reporting paper](<https://www.sota2.com/research/paper/69afa5662d603e6de3e43878.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa738e1e209131cea5d34e.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### DACA

Pornography ASR: 28. Result ID: 69afd6c06dbb7075b958300c.

[Reporting paper](<https://www.sota2.com/research/paper/69afa5662d603e6de3e43878.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### VEIL

Pornography ASR: 80. Result ID: 69afd6c06dbb7075b958300d.

[Reporting paper](<https://www.sota2.com/research/paper/69afa5662d603e6de3e43878.md>)

[Method origin paper](<https://www.sota2.com/research/paper/691bfe286750be83fe5ced41.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### TFM

Pornography ASR: 90. Result ID: 69afd6c06dbb7075b958300e.

[Reporting paper](<https://www.sota2.com/research/paper/69afa5662d603e6de3e43878.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69afa5662d603e6de3e43878.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

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