List your company

Indirect Prompt Injection on Multi-News

100ASRJudgeDeceiver
-4235077Dec 10, 2025

Evaluation results

MethodLinks
JudgeDeceiverTarget LLM=Qwen3-4B2025.12100—
ObliInjection-GCGTarget LLM=Llama4-17B2025.12100—
ObliInjectionTarget LLM=Llama-3-8B2025.12100—
ObliInjectionTarget LLM=Llama4-17B2025.12100—
ObliInjectionTarget LLM=Qwen3-4B2025.12100—
ObliInjection-CETarget LLM=Falcon3-7B2025.1299.8—
ObliInjection-GCGTarget LLM=Qwen3-4B2025.1299.6—
ObliInjection-CETarget LLM=Llama-3-8B2025.1299.6—
ObliInjection-CETarget LLM=Llama4-17B2025.1299.6—
ObliInjectionTarget LLM=Llama-3.1-8B2025.1299.6—
ObliInjectionTarget LLM=Qwen-2.5-7B2025.1299.6—
ObliInjection-CETarget LLM=Llama-3.1-8B2025.1299.4—
ObliInjectionTarget LLM=Falcon3-7B2025.1298—
Neural ExecTarget LLM=Llama4-17B2025.1297—
ObliInjection-GCGTarget LLM=Qwen-2.5-7B2025.1294.2—
JudgeDeceiverTarget LLM=Qwen-2.5-7B2025.1294—
ObliInjectionTarget LLM=Mistral-7B2025.1293.8—
ObliInjection-CETarget LLM=Qwen3-4B2025.1291—
Neural ExecTarget LLM=Qwen3-4B2025.1289—
Combined AttackTarget LLM=Qwen3-4B2025.1287.5—
ObliInjection-GCGTarget LLM=Llama-3-8B2025.1282—
JudgeDeceiverTarget LLM=Llama-3-8B2025.1276.6—
ObliInjection-CETarget LLM=Mistral-7B2025.1267—
Neural ExecTarget LLM=Llama-3-8B2025.1258.4—
Combined AttackTarget LLM=Llama4-17B2025.1251.6—
ObliInjection-CETarget LLM=Qwen-2.5-7B2025.1240.4—
ObliInjection-GCGTarget LLM=Llama-3.1-8B2025.1234.8—
ObliInjection-GCGTarget LLM=Falcon3-7B2025.1219.6—
Combined AttackTarget LLM=Falcon3-7B2025.128.4—
Neural ExecTarget LLM=Mistral-7B2025.127.8—
Combined AttackTarget LLM=Qwen-2.5-7B2025.127—
JudgeDeceiverTarget LLM=Falcon3-7B2025.125—
Neural ExecTarget LLM=Llama-3.1-8B2025.123—
JudgeDeceiverTarget LLM=Mistral-7B2025.122.6—
JudgeDeceiverTarget LLM=Llama4-17B2025.122.4—
Combined AttackTarget LLM=Llama-3.1-8B2025.120.6—
Neural ExecTarget LLM=Falcon3-7B2025.120.4—
Combined AttackTarget LLM=Llama-3-8B2025.120—
Combined AttackTarget LLM=Mistral-7B2025.120—
Neural ExecTarget LLM=Qwen-2.5-7B2025.120—
JudgeDeceiverTarget LLM=Llama-3.1-8B2025.120—
ObliInjection-GCGTarget LLM=Mistral-7B2025.120—
Combined AttackTarget LLM=Average2025.12—22.2
JudgeDeceiverTarget LLM=Average2025.12—40.1
Neural ExecTarget LLM=Average2025.12—36.5
ObliInjectionTarget LLM=Average2025.12—98.7
ObliInjection-CETarget LLM=Average2025.12—85.3
ObliInjection-GCGTarget LLM=Average2025.12—61.5