# Safety Alignment on VSA Violence (visual synonym attack prompts)

Source: [Safety Alignment on VSA Violence (visual synonym attack prompts)](<https://www.sota2.com/research/sota/safety-alignment-on-vsa-violence-visual-synonym-attack-prompts>)

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

Data updated: 2026-07-08T07:36:20.383Z

Evaluation of model safety alignment performance using the VSA Violence (visual synonym attack prompts) 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 | Safety Alignment |
| Dataset (including any stated split) | VSA Violence (visual synonym attack prompts) |
| Metric | ASR |
| Unit | PERCENT |
| Direction | Lower is better |
| Results loaded | 16 of 16 |
| 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/safety-alignment-on-vsa-violence-visual-synonym-attack-prompts.md?metric=ASR>): Lower is better; unit: PERCENT.

## Recorded results

| Method | ASR | Result date | Report paper |
| --- | --- | --- | --- |
| AEGIS | 0 | Not recorded | 6a4dd9904da6617fe9198b3d |
| AEGIS | 4 | Not recorded | 6a4dd9904da6617fe9198b3d |
| SafeClip | 5 | Not recorded | 6a4dd9904da6617fe9198b3d |
| SLD-Strong | 5 | Not recorded | 6a4dd9904da6617fe9198b3d |
| SAFREE | 8 | Not recorded | 6a4dd9904da6617fe9198b3d |
| SafeGuider | 11 | Not recorded | 6a4dd9904da6617fe9198b3d |
| CAD | 12 | Not recorded | 6a4dd9904da6617fe9198b3d |
| UCE | 15 | Not recorded | 6a4dd9904da6617fe9198b3d |
| Receler | 17 | Not recorded | 6a4dd9904da6617fe9198b3d |
| SDID | 21 | Not recorded | 6a4dd9904da6617fe9198b3d |
| ESD-u | 30 | Not recorded | 6a4dd9904da6617fe9198b3d |
| SLD-Medium | 32 | Not recorded | 6a4dd9904da6617fe9198b3d |
| Diff-QuickFix | 45 | Not recorded | 6a4dd9904da6617fe9198b3d |
| SD 2.1 | 51 | Not recorded | 6a4dd9904da6617fe9198b3d |
| SLD-Weak | 54 | Not recorded | 6a4dd9904da6617fe9198b3d |
| SD 1.4 | 65 | Not recorded | 6a4dd9904da6617fe9198b3d |

### AEGIS

ASR: 0. Result ID: 6a4dfde460b88bb339d741ac.

[Reporting paper](<https://www.sota2.com/research/paper/6a4dd9904da6617fe9198b3d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a4dd9904da6617fe9198b3d.md>)

Recorded evaluation configuration: {"Category":"AEGIS (Ours)","Base Model":"SD 1.4"}. This record does not assert independent reproduction.

### AEGIS

ASR: 4. Result ID: 6a4dfde460b88bb339d741ad.

[Reporting paper](<https://www.sota2.com/research/paper/6a4dd9904da6617fe9198b3d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a4dd9904da6617fe9198b3d.md>)

Recorded evaluation configuration: {"Category":"AEGIS (Ours)","Base Model":"SD 2.1"}. This record does not assert independent reproduction.

### SafeClip

ASR: 5. Result ID: 6a4dfde460b88bb339d741a0.

[Reporting paper](<https://www.sota2.com/research/paper/6a4dd9904da6617fe9198b3d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa74481e209131cea60c22.md>)

Recorded evaluation configuration: {"Category":"Input-Space"}. This record does not assert independent reproduction.

### SLD-Strong

ASR: 5. Result ID: 6a4dfde460b88bb339d741a9.

[Reporting paper](<https://www.sota2.com/research/paper/6a4dd9904da6617fe9198b3d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6f231e209131cea43f5d.md>)

[Implementation](<https://huggingface.co/docs/diffusers/api/pipelines/stable_diffusion_safe>)

Recorded evaluation configuration: {"Category":"Trigger-Specific"}. This record does not assert independent reproduction.

### SAFREE

ASR: 8. Result ID: 6a4dfde460b88bb339d741a1.

[Reporting paper](<https://www.sota2.com/research/paper/6a4dd9904da6617fe9198b3d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa77721e209131cea7326a.md>)

Recorded evaluation configuration: {"Category":"Input-Space"}. This record does not assert independent reproduction.

### SafeGuider

ASR: 11. Result ID: 6a4dfde460b88bb339d741a2.

[Reporting paper](<https://www.sota2.com/research/paper/6a4dd9904da6617fe9198b3d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7c1e1e209131cea8c7e4.md>)

Recorded evaluation configuration: {"Category":"Input-Space"}. This record does not assert independent reproduction.

### CAD

ASR: 12. Result ID: 6a4dfde460b88bb339d741ab.

[Reporting paper](<https://www.sota2.com/research/paper/6a4dd9904da6617fe9198b3d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa77641e209131cea73004.md>)

Recorded evaluation configuration: {"Category":"Feature Pruning"}. This record does not assert independent reproduction.

### UCE

ASR: 15. Result ID: 6a4dfde460b88bb339d741a4.

[Reporting paper](<https://www.sota2.com/research/paper/6a4dd9904da6617fe9198b3d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa755b1e209131cea67702.md>)

Recorded evaluation configuration: {"Category":"Trigger-Specific"}. This record does not assert independent reproduction.

### Receler

ASR: 17. Result ID: 6a4dfde460b88bb339d741a5.

[Reporting paper](<https://www.sota2.com/research/paper/6a4dd9904da6617fe9198b3d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa74241e209131cea5ff33.md>)

Recorded evaluation configuration: {"Category":"Trigger-Specific"}. This record does not assert independent reproduction.

### SDID

ASR: 21. Result ID: 6a4dfde460b88bb339d741aa.

[Reporting paper](<https://www.sota2.com/research/paper/6a4dd9904da6617fe9198b3d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa72ac1e209131cea57c2d.md>)

[Implementation](<https://interpretdiffusion.github.io/>)

Recorded evaluation configuration: {"Category":"Trigger-Specific"}. This record does not assert independent reproduction.

### ESD-u

ASR: 30. Result ID: 6a4dfde460b88bb339d741a3.

[Reporting paper](<https://www.sota2.com/research/paper/6a4dd9904da6617fe9198b3d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6fc61e209131cea47556.md>)

Recorded evaluation configuration: {"Category":"Trigger-Specific"}. This record does not assert independent reproduction.

### SLD-Medium

ASR: 32. Result ID: 6a4dfde460b88bb339d741a8.

[Reporting paper](<https://www.sota2.com/research/paper/6a4dd9904da6617fe9198b3d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6f231e209131cea43f5d.md>)

[Implementation](<https://huggingface.co/docs/diffusers/api/pipelines/stable_diffusion_safe>)

Recorded evaluation configuration: {"Category":"Trigger-Specific"}. This record does not assert independent reproduction.

### Diff-QuickFix

ASR: 45. Result ID: 6a4dfde460b88bb339d741a6.

[Reporting paper](<https://www.sota2.com/research/paper/6a4dd9904da6617fe9198b3d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa70f31e209131cea4e533.md>)

Recorded evaluation configuration: {"Category":"Trigger-Specific"}. This record does not assert independent reproduction.

### SD 2.1

ASR: 51. Result ID: 6a4dfde460b88bb339d7419f.

[Reporting paper](<https://www.sota2.com/research/paper/6a4dd9904da6617fe9198b3d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6c391e209131cea32e4a.md>)

[Implementation](<https://github.com/CompVis/latent-diffusion>)

Recorded evaluation configuration: {"Category":"No Alignment"}. This record does not assert independent reproduction.

### SLD-Weak

ASR: 54. Result ID: 6a4dfde460b88bb339d741a7.

[Reporting paper](<https://www.sota2.com/research/paper/6a4dd9904da6617fe9198b3d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6f231e209131cea43f5d.md>)

[Implementation](<https://huggingface.co/docs/diffusers/api/pipelines/stable_diffusion_safe>)

Recorded evaluation configuration: {"Category":"Trigger-Specific"}. This record does not assert independent reproduction.

### SD 1.4

ASR: 65. Result ID: 6a4dfde460b88bb339d7419e.

[Reporting paper](<https://www.sota2.com/research/paper/6a4dd9904da6617fe9198b3d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6c391e209131cea32e4a.md>)

[Implementation](<https://github.com/CompVis/latent-diffusion>)

Recorded evaluation configuration: {"Category":"No Alignment"}. This record does not assert independent reproduction.

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