List your company

Adversarial Attack against SeeAct agent on Mind2Web 600 tasks (test)

54.1ASR Finance (pass@10)Genesis
-1.43612.98227.441.818Oct 21, 2025

Evaluation results

MethodLinks
GenesisBackend LLM=GPT-4o2025.1054.155.343.958.753
GenesisBackend LLM=GPT-4o, Initialization Setting=w/o Initialization2025.1044.143.843.746.344.5
GenesisBackend LLM=Gemini-2.5-Flash2025.1043.94445.246.644.9
AdvAgentBackend LLM=GPT-4o2025.1042.843.544.14443.6
GenesisBackend LLM=Gemini-2.5-Flash, Initialization Setting=w/o Initialization2025.1034.935.736.237.536.1
AdvAgentBackend LLM=Gemini-2.5-Flash2025.1034.834.335.435.535
EIABackend LLM=GPT-4o2025.1034.234.335.535.334.8
EIABackend LLM=Gemini-2.5-Flash2025.1027.527.628.428.227.9
InjecAttackBackend LLM=GPT-4o2025.102425.325.426.525.3
GenesisBackend LLM=GPT-52025.102323.523.825.724
InjecAttackBackend LLM=Gemini-2.5-Flash2025.1019.220.420.521.120.3
GenesisBackend LLM=GPT-5, Initialization Setting=w/o Initialization2025.1019.118.719.719.819.3
AdvAgentBackend LLM=GPT-52025.1018.124.118.619.320
EIABackend LLM=GPT-52025.1014.814.615.215.114.9
AgentAttackBackend LLM=GPT-4o2025.1014.714.115.915.815.1
AgentAttackBackend LLM=Gemini-2.5-Flash2025.1011.811.412.812.712.2
InjecAttackBackend LLM=GPT-52025.1010.110.910.811.410.8
AgentAttackBackend LLM=GPT-52025.106.466.96.86.5
I-GCGBackend LLM=GPT-4o2025.103.53.74.24.64
I-GCGBackend LLM=Gemini-2.5-Flash2025.102.43.334.13.2
GCGBackend LLM=Gemini-2.5-Flash2025.101.31.52.12.41.8
I-GCGBackend LLM=GPT-52025.101.221.72.61.9
GCGBackend LLM=GPT-4o2025.101.12.52.23.42.3
GCGBackend LLM=GPT-52025.100.70.91.51.81.2