Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Jailbreak Attack on Harmful Query Evaluation Set N=750 GPT-5.4-nano

1.05Toxicity Score

HADES

0.93081.73542.543.3446May 7, 2026
Updated 26d ago

Evaluation Results

MethodLinks
2026.05
1.050.881.331.73
2026.05
1.111.233.23.33
2026.05
1.160.724.85.6
2026.05
1.171.674.275.07
2026.05
1.862.7525.7314.4
2026.05
2.624.2642.833.73
2026.05
3.385.6461.0754
2026.05
3.656.1569.662.4
2026.05
3.756.4271.4768.93
2026.05
4.036.9478.6774.67