List your company

Video Jailbreaking on MM-SafetyBench 1.0 (test)

96Attack Success RateRunaway Evil
77.2882.148791.86Dec 7, 2025

Evaluation results

MethodLinks
Runaway EvilVideo Generation Model=Opensora, Evaluation Model (MLLM)=LLAVA2025.1296
Runaway EvilVideo Generation Model=Wan, Evaluation Model (MLLM)=LLAVA2025.1295
Runaway EvilVideo Generation Model=DynamiCrafter, Evaluation Model (MLLM)=LLAVA2025.1295
Runaway EvilVideo Generation Model=Cogvideo, Evaluation Model (MLLM)=LLAVA2025.1295
Runaway EvilVideo Generation Model=Wan, Evaluation Model (MLLM)=GEMMA2025.1294
Runaway EvilVideo Generation Model=Opensora, Evaluation Model (MLLM)=GEMMA2025.1294
Runaway EvilVideo Generation Model=Cogvideo, Evaluation Model (MLLM)=GEMMA2025.1294
Runaway EvilVideo Generation Model=DynamiCrafter, Evaluation Model (MLLM)=GEMMA2025.1293
Runaway EvilVideo Generation Model=Wan, Evaluation Model (MLLM)=QWEN2025.1290
Runaway EvilVideo Generation Model=Opensora, Evaluation Model (MLLM)=QWEN2025.1290
Runaway EvilVideo Generation Model=DynamiCrafter, Evaluation Model (MLLM)=QWEN2025.1288
Runaway EvilVideo Generation Model=Cogvideo, Evaluation Model (MLLM)=QWEN2025.1288
SneakyVideo Generation Model=Opensora, Evaluation Model (MLLM)=GEMMA2025.1287
PGJVideo Generation Model=Cogvideo, Evaluation Model (MLLM)=LLAVA2025.1287
SneakyVideo Generation Model=DynamiCrafter, Evaluation Model (MLLM)=GEMMA2025.1286
SneakyVideo Generation Model=Opensora, Evaluation Model (MLLM)=LLAVA2025.1286
SneakyVideo Generation Model=Cogvideo, Evaluation Model (MLLM)=LLAVA2025.1286
PGJVideo Generation Model=Wan, Evaluation Model (MLLM)=LLAVA2025.1286
PGJVideo Generation Model=Opensora, Evaluation Model (MLLM)=LLAVA2025.1286
PGJVideo Generation Model=Cogvideo, Evaluation Model (MLLM)=GEMMA2025.1286
DACAVideo Generation Model=Cogvideo, Evaluation Model (MLLM)=LLAVA2025.1286
DACAVideo Generation Model=Cogvideo, Evaluation Model (MLLM)=GEMMA2025.1286
SneakyVideo Generation Model=DynamiCrafter, Evaluation Model (MLLM)=LLAVA2025.1285
PGJVideo Generation Model=DynamiCrafter, Evaluation Model (MLLM)=GEMMA2025.1285
DACAVideo Generation Model=DynamiCrafter, Evaluation Model (MLLM)=LLAVA2025.1285
DACAVideo Generation Model=Opensora, Evaluation Model (MLLM)=GEMMA2025.1285
SneakyVideo Generation Model=Wan, Evaluation Model (MLLM)=LLAVA2025.1284
SneakyVideo Generation Model=Cogvideo, Evaluation Model (MLLM)=GEMMA2025.1284
PGJVideo Generation Model=Opensora, Evaluation Model (MLLM)=GEMMA2025.1284
DACAVideo Generation Model=Wan, Evaluation Model (MLLM)=LLAVA2025.1284
DACAVideo Generation Model=Opensora, Evaluation Model (MLLM)=LLAVA2025.1284
PGJVideo Generation Model=Wan, Evaluation Model (MLLM)=GEMMA2025.1283
PGJVideo Generation Model=DynamiCrafter, Evaluation Model (MLLM)=LLAVA2025.1283
DACAVideo Generation Model=Wan, Evaluation Model (MLLM)=GEMMA2025.1283
SneakyVideo Generation Model=Cogvideo, Evaluation Model (MLLM)=QWEN2025.1282
PGJVideo Generation Model=Opensora, Evaluation Model (MLLM)=QWEN2025.1282
DACAVideo Generation Model=DynamiCrafter, Evaluation Model (MLLM)=GEMMA2025.1282
SneakyVideo Generation Model=Wan, Evaluation Model (MLLM)=GEMMA2025.1281
PGJVideo Generation Model=Wan, Evaluation Model (MLLM)=QWEN2025.1281
PGJVideo Generation Model=Cogvideo, Evaluation Model (MLLM)=QWEN2025.1281
DACAVideo Generation Model=DynamiCrafter, Evaluation Model (MLLM)=QWEN2025.1281
DACAVideo Generation Model=Cogvideo, Evaluation Model (MLLM)=QWEN2025.1281
PGJVideo Generation Model=DynamiCrafter, Evaluation Model (MLLM)=QWEN2025.1280
SneakyVideo Generation Model=Opensora, Evaluation Model (MLLM)=QWEN2025.1279
DACAVideo Generation Model=Wan, Evaluation Model (MLLM)=QWEN2025.1279
SneakyVideo Generation Model=Wan, Evaluation Model (MLLM)=QWEN2025.1278
SneakyVideo Generation Model=DynamiCrafter, Evaluation Model (MLLM)=QWEN2025.1278
DACAVideo Generation Model=Opensora, Evaluation Model (MLLM)=QWEN2025.1278