# Multimodal Attack Detection on Multimodal attack detection (test)

Source: [Multimodal Attack Detection on Multimodal attack detection (test)](<https://www.sota2.com/research/sota/multimodal-attack-detection-on-multimodal-attack-detection-test>)

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

Data updated: 2026-04-09T23:17:57.111Z

Evaluation of model performance for Multimodal Attack Detection using the specified test split of the Multimodal attack detection 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 | Multimodal Attack Detection |
| Dataset (including any stated split) | Multimodal attack detection (test) |
| Metric | Precision |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 10 of 10 |
| 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

- [Precision](<https://www.sota2.com/research/sota/multimodal-attack-detection-on-multimodal-attack-detection-test.md?metric=Precision>): Higher is better; unit: PERCENT.

- [Recall](<https://www.sota2.com/research/sota/multimodal-attack-detection-on-multimodal-attack-detection-test.md?metric=Recall>): Higher is better; unit: PERCENT.

- [F1 Score](<https://www.sota2.com/research/sota/multimodal-attack-detection-on-multimodal-attack-detection-test.md?metric=F1%20Score>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Precision | Result date | Report paper |
| --- | --- | --- | --- |
| Gemini-2.5-Flash-Lite† | 100 | Not recorded | 69d732667d0172140ec7698b |
| GPT-4.1 Mini† | 100 | Not recorded | 69d732667d0172140ec7698b |
| SALLIE | 99 | Not recorded | 69d732667d0172140ec7698b |
| HiddenDetect | 97 | Not recorded | 69d732667d0172140ec7698b |
| SALLIE | 96 | Not recorded | 69d732667d0172140ec7698b |
| SALLIE | 94 | Not recorded | 69d732667d0172140ec7698b |
| JailGuard | 87 | Not recorded | 69d732667d0172140ec7698b |
| PIShield | 57 | Not recorded | 69d732667d0172140ec7698b |
| EEG-Defender | 56 | Not recorded | 69d732667d0172140ec7698b |
| Perplexity Filter | 50 | Not recorded | 69d732667d0172140ec7698b |

### Gemini-2.5-Flash-Lite†

Precision: 100. Result ID: 69d8321c47f34ce924bfd937.

[Reporting paper](<https://www.sota2.com/research/paper/69d732667d0172140ec7698b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7c311e209131cea8cef5.md>)

Recorded evaluation configuration: {"Proprietary":"true"}. This record does not assert independent reproduction.

### GPT-4.1 Mini†

Precision: 100. Result ID: 69d8321c47f34ce924bfd938.

[Reporting paper](<https://www.sota2.com/research/paper/69d732667d0172140ec7698b.md>)

Recorded evaluation configuration: {"Proprietary":"true"}. This record does not assert independent reproduction.

### SALLIE

Precision: 99. Result ID: 69d8321c47f34ce924bfd934.

[Reporting paper](<https://www.sota2.com/research/paper/69d732667d0172140ec7698b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69d732667d0172140ec7698b.md>)

Recorded evaluation configuration: {"Backbone":"SmolVLM2-2.2B-Instruct"}. This record does not assert independent reproduction.

### HiddenDetect

Precision: 97. Result ID: 69d8321c47f34ce924bfd932.

[Reporting paper](<https://www.sota2.com/research/paper/69d732667d0172140ec7698b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa79951e209131cea7e65c.md>)

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

### SALLIE

Precision: 96. Result ID: 69d8321c47f34ce924bfd935.

[Reporting paper](<https://www.sota2.com/research/paper/69d732667d0172140ec7698b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69d732667d0172140ec7698b.md>)

Recorded evaluation configuration: {"Backbone":"Phi-3.5-vision-instruct"}. This record does not assert independent reproduction.

### SALLIE

Precision: 94. Result ID: 69d8321c47f34ce924bfd936.

[Reporting paper](<https://www.sota2.com/research/paper/69d732667d0172140ec7698b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69d732667d0172140ec7698b.md>)

Recorded evaluation configuration: {"Backbone":"Gemma-3-4b-it"}. This record does not assert independent reproduction.

### JailGuard

Precision: 87. Result ID: 69d8321c47f34ce924bfd933.

[Reporting paper](<https://www.sota2.com/research/paper/69d732667d0172140ec7698b.md>)

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

### PIShield

Precision: 57. Result ID: 69d8321c47f34ce924bfd931.

[Reporting paper](<https://www.sota2.com/research/paper/69d732667d0172140ec7698b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7c451e209131cea8d4c4.md>)

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

### EEG-Defender

Precision: 56. Result ID: 69d8321c47f34ce924bfd930.

[Reporting paper](<https://www.sota2.com/research/paper/69d732667d0172140ec7698b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7acc1e209131cea85266.md>)

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

### Perplexity Filter

Precision: 50. Result ID: 69d8321c47f34ce924bfd92f.

[Reporting paper](<https://www.sota2.com/research/paper/69d732667d0172140ec7698b.md>)

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