# Morality Attack (Defense on User Input) on Designed Morality Attacks 1.0 (test)

Source: [Morality Attack (Defense on User Input) on Designed Morality Attacks 1.0 (test)](<https://www.sota2.com/research/sota/morality-attack-defense-on-user-input-on-designed-morality-attacks-1-0-test>)

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

Data updated: 2026-04-21T09:45:25.438Z

Evaluation of model performance against designed morality attacks on the Designed Morality Attacks 1.0 dataset (test split), reporting metrics such as RN, VN, FN, BN, Avg, and Needs Caution Rate.

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 | Morality Attack (Defense on User Input) |
| Dataset (including any stated split) | Designed Morality Attacks 1.0 (test) |
| Metric | RN |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 7 of 7 |
| 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

- [RN](<https://www.sota2.com/research/sota/morality-attack-defense-on-user-input-on-designed-morality-attacks-1-0-test.md?metric=RN>): Higher is better; unit: PERCENT.

- [VN](<https://www.sota2.com/research/sota/morality-attack-defense-on-user-input-on-designed-morality-attacks-1-0-test.md?metric=VN>): Higher is better; unit: PERCENT.

- [FN](<https://www.sota2.com/research/sota/morality-attack-defense-on-user-input-on-designed-morality-attacks-1-0-test.md?metric=FN>): Higher is better; unit: PERCENT.

- [BN](<https://www.sota2.com/research/sota/morality-attack-defense-on-user-input-on-designed-morality-attacks-1-0-test.md?metric=BN>): Higher is better; unit: PERCENT.

- [Average Score (Avg)](<https://www.sota2.com/research/sota/morality-attack-defense-on-user-input-on-designed-morality-attacks-1-0-test.md?metric=Average%20Score%20(Avg)>): Higher is better; unit: PERCENT.

- [Needs Caution Rate](<https://www.sota2.com/research/sota/morality-attack-defense-on-user-input-on-designed-morality-attacks-1-0-test.md?metric=Needs%20Caution%20Rate>): Lower is better; unit: PERCENT.

## Recorded results

| Method | RN | Result date | Report paper |
| --- | --- | --- | --- |
| ShieldGemma-9B | 100 | Not recorded | 69e704c37d0172140ec7eef6 |
| Llama-Guard-4 | 99.9 | Not recorded | 69e704c37d0172140ec7eef6 |
| WildGuard | 96.8 | Not recorded | 69e704c37d0172140ec7eef6 |
| Prompt-Guard-2 (86M) | 93.2 | Not recorded | 69e704c37d0172140ec7eef6 |
| Granite-Guardian-3.3-8B | 92.8 | Not recorded | 69e704c37d0172140ec7eef6 |
| Aegis Permissive (CP) | 65.3 | Not recorded | 69e704c37d0172140ec7eef6 |
| Aegis Defensive (CP) | 16 | Not recorded | 69e704c37d0172140ec7eef6 |

### ShieldGemma-9B

RN: 100. Result ID: 69e7458f47f34ce924cde543.

[Reporting paper](<https://www.sota2.com/research/paper/69e704c37d0172140ec7eef6.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa74661e209131cea617ba.md>)

Recorded evaluation configuration: {"Justification":"false"}. This record does not assert independent reproduction.

### Llama-Guard-4

RN: 99.9. Result ID: 69e7458f47f34ce924cde540.

[Reporting paper](<https://www.sota2.com/research/paper/69e704c37d0172140ec7eef6.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa75c21e209131cea69c6e.md>)

Recorded evaluation configuration: {"Justification":"false"}. This record does not assert independent reproduction.

### WildGuard

RN: 96.8. Result ID: 69e7458f47f34ce924cde53f.

[Reporting paper](<https://www.sota2.com/research/paper/69e704c37d0172140ec7eef6.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa76291e209131cea6b9a6.md>)

Recorded evaluation configuration: {"Justification":"false"}. This record does not assert independent reproduction.

### Prompt-Guard-2 (86M)

RN: 93.2. Result ID: 69e7458f47f34ce924cde542.

[Reporting paper](<https://www.sota2.com/research/paper/69e704c37d0172140ec7eef6.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6aef1e209131cea2b2a6.md>)

[Implementation](<https://github.com/microsoft/DeBERTa111>)

Recorded evaluation configuration: {"Justification":"false"}. This record does not assert independent reproduction.

### Granite-Guardian-3.3-8B

RN: 92.8. Result ID: 69e7458f47f34ce924cde541.

[Reporting paper](<https://www.sota2.com/research/paper/69e704c37d0172140ec7eef6.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa76331e209131cea6bcd3.md>)

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

### Aegis Permissive (CP)

RN: 65.3. Result ID: 69e7458f47f34ce924cde53e.

[Reporting paper](<https://www.sota2.com/research/paper/69e704c37d0172140ec7eef6.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa74ca1e209131cea63d99.md>)

Recorded evaluation configuration: {"Justification":"false"}. This record does not assert independent reproduction.

### Aegis Defensive (CP)

RN: 16. Result ID: 69e7458f47f34ce924cde53d.

[Reporting paper](<https://www.sota2.com/research/paper/69e704c37d0172140ec7eef6.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa74ca1e209131cea63d99.md>)

Recorded evaluation configuration: {"Justification":"false"}. This record does not assert independent reproduction.

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