List your company

System Prompt Extraction on Attack Prompts In-domain

78Attack Success Rate (TA)RogueMerge
6.2424.8743.562.13Jun 2, 2026

Evaluation results

MethodLinks
RogueMerge2026.067880767878
LoBAM*2026.063446453439.8
MergeHijacking*2026.061617151716.3
SFT2026.06101371411
Clean2026.0697877.8