Prompt Injection on CSQA
18.33ASR@3G-Safeguard
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| G-SafeguardTopology=Tree2025.12 | 18.33 | 100 | |
| G-SafeguardTopology=Star2025.12 | 18.33 | 100 | |
| G-SafeguardTopology=Random2025.12 | 18.67 | 98.22 | |
| G-SafeguardTopology=Tree, Backbone LLM=GPT-4o-mini2025.08 | 18.67 | 99.56 | |
| G-SafeguardTopology=Random, Backbone LLM=GPT-4o-mini2025.08 | 19 | 99.11 | |
| G-SafeguardTopology=Chain, Backbone LLM=GPT-4o-mini2025.08 | 19.33 | 100 | |
| G-SafeguardTopology=Star, Backbone LLM=GPT-4o-mini2025.08 | 19.33 | 100 | |
| G-SafeguardTopology=Chain2025.12 | 19.67 | 100 | |
| XG-GuardTopology=Star2025.12 | 20.67 | 91.11 | |
| XG-GuardTopology=Chain2025.12 | 21.67 | 87.11 | |
| BlindGuardTopology=Star, Backbone LLM=GPT-4o-mini2025.08 | 22 | 85.78 | |
| XG-GuardTopology=Tree2025.12 | 22.66 | 89.78 | |
| PREMTopology=Tree, Backbone LLM=GPT-4o-mini2025.08 | 22.67 | 50.67 | |
| BlindGuardTopology=Star2025.12 | 23.33 | 83.56 | |
| BlindGuardTopology=Chain2025.12 | 23.67 | 77.78 | |
| TAMTopology=Tree2025.12 | 23.67 | 55.56 | |
| BlindGuardTopology=Chain, Backbone LLM=GPT-4o-mini2025.08 | 23.67 | 80 | |
| BlindGuardTopology=Tree, Backbone LLM=GPT-4o-mini2025.08 | 24 | 74.67 | |
| PREMTopology=Chain2025.12 | 25 | 54.22 | |
| BlindGuardTopology=Tree2025.12 | 25 | 75.11 | |
| BlindGuardTopology=Random2025.12 | 25 | 75.11 | |
| XG-GuardTopology=Random2025.12 | 25 | 90.67 | |
| DOMINANTTopology=Tree2025.12 | 25.67 | 46.67 | |
| TAMTopology=Tree, Backbone LLM=GPT-4o-mini2025.08 | 26 | 53.78 | |
| DOMINANTTopology=Chain2025.12 | 26.33 | 44.89 | |
| PREMTopology=Tree2025.12 | 26.33 | 52 | |
| DOMINANTTopology=Tree, Backbone LLM=GPT-4o-mini2025.08 | 27 | 44.89 | |
| TAMTopology=Chain, Backbone LLM=GPT-4o-mini2025.08 | 27.33 | 55.11 | |
| TAMTopology=Chain2025.12 | 28 | 54.67 | |
| TAMTopology=Star, Backbone LLM=GPT-4o-mini2025.08 | 28.67 | 57.33 | |
| PREMTopology=Chain, Backbone LLM=GPT-4o-mini2025.08 | 29.33 | 50.67 | |
| BlindGuardTopology=Random, Backbone LLM=GPT-4o-mini2025.08 | 30 | 75.11 | |
| PREMTopology=Star2025.12 | 30.33 | 50.67 | |
| DOMINANTTopology=Chain, Backbone LLM=GPT-4o-mini2025.08 | 30.33 | 47.11 | |
| PREMTopology=Random2025.12 | 31.33 | 53.78 | |
| TAMTopology=Random2025.12 | 31.33 | 44.44 | |
| No DefenseTopology=Tree2025.12 | 32.67 | — | |
| No DefenseTopology=Tree, Backbone LLM=GPT-4o-mini2025.08 | 33 | — | |
| DOMINANTTopology=Random2025.12 | 34 | 45.33 | |
| TAMTopology=Star2025.12 | 34.33 | 56.44 | |
| PREMTopology=Star, Backbone LLM=GPT-4o-mini2025.08 | 34.33 | 56.89 | |
| PREMTopology=Random, Backbone LLM=GPT-4o-mini2025.08 | 34.67 | 52.44 | |
| TAMTopology=Random, Backbone LLM=GPT-4o-mini2025.08 | 35.33 | 45.33 | |
| DOMINANTTopology=Random, Backbone LLM=GPT-4o-mini2025.08 | 37 | 46.67 | |
| No DefenseTopology=Random2025.12 | 38 | — | |
| DOMINANTTopology=Star, Backbone LLM=GPT-4o-mini2025.08 | 38.67 | 38.22 | |
| DOMINANTTopology=Star2025.12 | 40.33 | 44 | |
| No DefenseTopology=Random, Backbone LLM=GPT-4o-mini2025.08 | 42 | — | |
| No DefenseTopology=Chain, Backbone LLM=GPT-4o-mini2025.08 | 42.33 | — | |
| No DefenseTopology=Chain2025.12 | 43 | — | |
| No DefenseTopology=Star2025.12 | 47 | — | |
| No DefenseTopology=Star, Backbone LLM=GPT-4o-mini2025.08 | 50.33 | — |