Jailbreak Attack on Multimodal Jailbreak Dataset Single Image (test)
0ASRClaude 3.5 Sonnet
Evaluation Results
| Method | Links | |
|---|---|---|
| Claude 3.5 SonnetAttack Type=Unperturbed2025.04 | 0 | |
| GPT-4o MiniAttack Type=Perturbed2025.04 | 0 | |
| Claude 3.5 SonnetAttack Type=Perturbed2025.04 | 1 | |
| GPT-4o MiniAttack Type=Unperturbed2025.04 | 3 | |
| GPT-4oAttack Type=Perturbed2025.04 | 4 | |
| Claude 3 HaikuAttack Type=Cipher2025.04 | 6 | |
| Claude 3.5 SonnetAttack Type=Cipher2025.04 | 9 | |
| Claude 3 HaikuAttack Type=Unperturbed2025.04 | 10 | |
| Gemini 1.5 ProAttack Type=Cipher2025.04 | 11 | |
| GPT-4oAttack Type=Unperturbed2025.04 | 13 | |
| Claude 3 HaikuAttack Type=Perturbed2025.04 | 17 | |
| GPT-4o MiniAttack Type=Cipher2025.04 | 21 | |
| Gemini 1.5 ProAttack Type=Unperturbed2025.04 | 23 | |
| GPT-4oAttack Type=Cipher2025.04 | 24 | |
| Gemini 1.5 ProAttack Type=Perturbed2025.04 | 26 | |
| Gemini 1.5 FlashAttack Type=Perturbed2025.04 | 28 | |
| Gemini 1.5 FlashAttack Type=Cipher2025.04 | 42 | |
| Gemini 1.5 FlashAttack Type=Unperturbed2025.04 | 72 |