Document Summarization on MultiNews
87ASRNo Defense
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| No DefenseAttack Type=Combined, Target LLM=Qwen-3-4B-Instruct-25072026.04 | 87 | — | |
| No DefenseAttack Type=Direct, Target LLM=Qwen-3-4B-Instruct-25072026.04 | 81 | — | |
| PromptArmorAttack Type=Direct, Target LLM=Qwen-3-4B-Instruct-25072026.04 | 78 | — | |
| PromptArmorAttack Type=Combined, Target LLM=Qwen-3-4B-Instruct-25072026.04 | 76 | — | |
| PromptGuardAttack Type=Direct, Target LLM=Qwen-3-4B-Instruct-25072026.04 | 66 | — | |
| PIGuardAttack Type=Direct, Target LLM=Qwen-3-4B-Instruct-25072026.04 | 55 | — | |
| PromptGuardAttack Type=Combined, Target LLM=Qwen-3-4B-Instruct-25072026.04 | 47 | — | |
| PIGuardAttack Type=Combined, Target LLM=Qwen-3-4B-Instruct-25072026.04 | 37 | — | |
| DataSentinelAttack Type=Direct, Target LLM=Qwen-3-4B-Instruct-25072026.04 | 20 | — | |
| GPT-OSS-SafeguardAttack Type=Direct, Target LLM=Qwen-3-4B-Instruct-25072026.04 | 20 | — | |
| AgentWatcherAttack Type=Direct, Target LLM=Qwen-3-4B-Instruct-25072026.04 | 5 | — | |
| DataSentinelAttack Type=Combined, Target LLM=Qwen-3-4B-Instruct-25072026.04 | 2 | — | |
| GPT-OSS-SafeguardAttack Type=Combined, Target LLM=Qwen-3-4B-Instruct-25072026.04 | 0 | — | |
| AgentWatcherAttack Type=Combined, Target LLM=Qwen-3-4B-Instruct-25072026.04 | 0 | — | |
| AgentWatcherAttack Type=Clean, Target LLM=Qwen-3-4B-Instruct-25072026.04 | — | 25 | |
| DataSentinelAttack Type=Clean, Target LLM=Qwen-3-4B-Instruct-25072026.04 | — | 20 | |
| GPT-OSS-SafeguardAttack Type=Clean, Target LLM=Qwen-3-4B-Instruct-25072026.04 | — | 25 | |
| No DefenseAttack Type=Clean, Target LLM=Qwen-3-4B-Instruct-25072026.04 | — | 25 | |
| PIGuardAttack Type=Clean, Target LLM=Qwen-3-4B-Instruct-25072026.04 | — | 23 | |
| PromptArmorAttack Type=Clean, Target LLM=Qwen-3-4B-Instruct-25072026.04 | — | 24 | |
| PromptGuardAttack Type=Clean, Target LLM=Qwen-3-4B-Instruct-25072026.04 | — | 25 |