Group Collusive Attack Detection on MMLU
92Detection AccuracyGroupGuard
Evaluation Results
| Method | Links | |
|---|---|---|
| GroupGuardAttack Type=Group Exclusion, Model=Deepseek-V3.12026.03 | 92 | |
| GroupGuardAttack Type=Targeted Suppression, Model=Deepseek-V3.12026.03 | 89 | |
| G-safeguardAttack Type=Group Exclusion, Model=Deepseek-V3.12026.03 | 87.2 | |
| G-safeguardAttack Type=Targeted Suppression, Model=Deepseek-V3.12026.03 | 86.7 | |
| GroupGuardAttack Type=Targeted Suppression, Model=Qwen3-235B-A22B2026.03 | 84 | |
| G-safeguardAttack Type=False Consensus, Model=Deepseek-V3.12026.03 | 83.6 | |
| GroupGuardAttack Type=False Consensus, Model=Deepseek-V3.12026.03 | 83 | |
| G-safeguardAttack Type=Targeted Suppression, Model=Qwen3-235B-A22B2026.03 | 81 | |
| G-safeguardAttack Type=False Consensus, Model=Qwen3-32B2026.03 | 80.2 | |
| G-safeguardAttack Type=Group Exclusion, Model=Qwen3-235B-A22B2026.03 | 78.6 | |
| G-safeguardAttack Type=Group Exclusion, Model=Qwen3-32B2026.03 | 76.2 | |
| AutoDefenseAttack Type=Targeted Suppression, Model=Deepseek-V3.12026.03 | 75.8 | |
| GroupGuardAttack Type=False Consensus, Model=Qwen3-32B2026.03 | 75 | |
| GroupGuardAttack Type=Group Exclusion, Model=Qwen3-32B2026.03 | 72 | |
| G-safeguardAttack Type=Targeted Suppression, Model=Qwen3-32B2026.03 | 71.8 | |
| GroupGuardAttack Type=Group Exclusion, Model=Qwen3-235B-A22B2026.03 | 71 | |
| AutoDefenseAttack Type=False Consensus, Model=Deepseek-V3.12026.03 | 69.8 | |
| GroupGuardAttack Type=Targeted Suppression, Model=Qwen3-32B2026.03 | 69 | |
| AutoDefenseAttack Type=Group Exclusion, Model=Qwen3-235B-A22B2026.03 | 68.6 | |
| AutoDefenseAttack Type=Group Exclusion, Model=Qwen3-32B2026.03 | 67.6 | |
| G-safeguardAttack Type=False Consensus, Model=Qwen3-235B-A22B2026.03 | 66.2 | |
| AutoDefenseAttack Type=False Consensus, Model=Qwen3-235B-A22B2026.03 | 65.6 | |
| AutoDefenseAttack Type=False Consensus, Model=Qwen3-32B2026.03 | 64.8 | |
| AutoDefenseAttack Type=Group Exclusion, Model=Deepseek-V3.12026.03 | 62.8 | |
| GroupGuardAttack Type=False Consensus, Model=Qwen3-235B-A22B2026.03 | 62 | |
| AutoDefenseAttack Type=Targeted Suppression, Model=Qwen3-235B-A22B2026.03 | 60.2 | |
| AutoDefenseAttack Type=Targeted Suppression, Model=Qwen3-32B2026.03 | 49.8 |