Hijacking on ACIARENA Code
57.5UAAgentVerse + SANDWICH
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| AgentVerse + SANDWICHMAS Platform=AgentVerse, Defense=SANDWICH2026.04 | 57.5 | 32.5 | |
| AgentVerseMAS Platform=AgentVerse, Defense=None2026.04 | 52.78 | 50.28 | |
| AgentVerse + BERT DETECTORMAS Platform=AgentVerse, Defense=BERT DETECTOR2026.04 | 52.22 | 53.06 | |
| MetaGPT + SANDWICHMAS Platform=MetaGPT, Defense=SANDWICH2026.04 | 51.94 | 10 | |
| AgentVerse + DELIMITERMAS Platform=AgentVerse, Defense=DELIMITER2026.04 | 51.39 | 49.72 | |
| AgentVerse + G-SAFEGUARDMAS Platform=AgentVerse, Defense=G-SAFEGUARD2026.04 | 46.39 | 33.06 | |
| AutoGen + SANDWICHMAS Platform=AutoGen, Defense=SANDWICH2026.04 | 44.72 | 79.72 | |
| AgentVerse + AGRAILMAS Platform=AgentVerse, Defense=AGRAIL2026.04 | 44.17 | 40.28 | |
| Self ConsistencyBase LLM=GPT-4o-mini2026.04 | 43.89 | 95 | |
| MetaGPT + ACI-SENTINELMAS Platform=MetaGPT, Defense=ACI-SENTINEL2026.04 | 42.5 | 0 | |
| AgentVerseBase LLM=GPT-4o-mini2026.04 | 41.11 | 48.05 | |
| AgentVerse + ACI-SENTINELMAS Platform=AgentVerse, Defense=ACI-SENTINEL2026.04 | 36.67 | 11.67 | |
| AutoGen + ACI-SENTINELMAS Platform=AutoGen, Defense=ACI-SENTINEL2026.04 | 34.44 | 8.06 | |
| LLM DebateBase LLM=GPT-4o-mini2026.04 | 34.22 | 100 | |
| MetaGPTMAS Platform=MetaGPT, Defense=None2026.04 | 30.56 | 79.44 | |
| AutoGen + BERT DETECTORMAS Platform=AutoGen, Defense=BERT DETECTOR2026.04 | 30 | 96.39 | |
| AutoGen + DELIMITERMAS Platform=AutoGen, Defense=DELIMITER2026.04 | 28.61 | 95.56 | |
| MetaGPT + G-SAFEGUARDMAS Platform=MetaGPT, Defense=G-SAFEGUARD2026.04 | 27.78 | 46.39 | |
| AutoGenMAS Platform=AutoGen, Defense=None2026.04 | 27.5 | 92.78 | |
| AutoGenBase LLM=GPT-4o-mini2026.04 | 25 | 80.83 | |
| MetaGPTBase LLM=GPT-4o-mini2026.04 | 23.05 | 100 | |
| AutoGen + G-SAFEGUARDMAS Platform=AutoGen, Defense=G-SAFEGUARD2026.04 | 15.56 | 67.22 | |
| MetaGPT + BERT DETECTORMAS Platform=MetaGPT, Defense=BERT DETECTOR2026.04 | 12.78 | 91.11 | |
| MetaGPT + DELIMITERMAS Platform=MetaGPT, Defense=DELIMITER2026.04 | 12.22 | 67.22 | |
| CAMELBase LLM=GPT-4o-mini2026.04 | 7.78 | 20.28 | |
| AutoGen + AGRAILMAS Platform=AutoGen, Defense=AGRAIL2026.04 | 7.5 | 35.56 | |
| MetaGPT + AGRAILMAS Platform=MetaGPT, Defense=AGRAIL2026.04 | 0.28 | 11.67 |