List your company

Jailbreak Defense on JBB-Behaviors

0ASRSAGE
-1.8810.8123.536.19May 17, 2025Jun 22, 2025Jul 29, 2025Sep 4, 2025Oct 10, 2025Nov 16, 2025Dec 23, 2025

Evaluation results

MethodLinks
SAGEModel=Gemma22025.0501
ICDModel=Qwen2.52025.0501
Goal PrioritizationModel=Qwen2.52025.0501
IAModel=Qwen2.52025.0501
SAGEModel=Qwen2.52025.0501
ICDModel=Llama3.12025.0501
SAGEModel=Llama3.12025.0501
VAESystem=Gemini-2.0-pro2025.120—
UMKSystem=Gemini-2.0-pro2025.120—
VAESystem=Gemini-2.0-flash2025.120—
CipherChatSystem=Gemini-2.0-flash2025.120—
VAESystem=Gemini-2.0-pro2025.120—
UMKSystem=Gemini-2.0-pro2025.120—
Self-ExaminationModel=Gemma22025.0511.1
CipherChatSystem=GPT-4o2025.121—
HADESSystem=Gemini-2.0-pro2025.121—
ArtPromptSystem=Gemini-2.0-pro2025.121—
UMKSystem=Gemini-2.0-flash2025.121—
CipherChatSystem=GPT-4o2025.121—
UMKSystem=Gemini-2.0-flash2025.121—
ICDModel=Gemma22025.0521.13
Self-ExaminationModel=Qwen2.52025.0521
IAModel=Llama3.12025.0521
MM-SafetyBenchSystem=GPT-4o2025.122—
HIMRDSystem=GPT-4o2025.122—
VAESystem=Gemini-2.0-flash2025.122—
CipherChatSystem=Gemini-2.0-flash2025.122—
Self-ReminderModel=Gemma22025.0531.18
Goal PrioritizationModel=Gemma22025.0531
HIMRDSystem=GPT-4o2025.123—
FigStep-proSystem=Gemini-2.0-pro2025.123—
FigStep-proSystem=Gemini-2.0-pro2025.123—
IAModel=Gemma22025.0541
Self-ReminderModel=Qwen2.52025.0541.1
Self-ReminderModel=Llama3.12025.0541
Goal PrioritizationModel=Llama3.12025.0541
MM-SafetyBenchSystem=Grok-32025.124—
HADESSystem=Gemini-2.0-pro2025.124—
ArtPromptSystem=Gemini-2.0-pro2025.124—
Dream-v0-Instruct-7B + Self-reminder + DIFFUGUARDBase Model=Dream-v0-Instruct-7B, Defense Method=Self-reminder + DIFFUGUARD2025.094—
MM-SafetyBenchSystem=Gemini-2.0-pro2025.125—
MM-SafetyBenchSystem=Gemini-2.0-flash2025.125—
Dream-v0-Instruct-7B + DIFFUGUARDBase Model=Dream-v0-Instruct-7B, Defense Method=DIFFUGUARD2025.095—
UMKSystem=GPT-4o2025.126—
Dream-v0-Instruct-7B + Self-reminderBase Model=Dream-v0-Instruct-7B, Defense Method=Self-reminder2025.096—
Dream-v0-Instruct-7B + PPL-FilterBase Model=Dream-v0-Instruct-7B, Defense Method=PPL-Filter2025.096.67—
VAESystem=GPT-4o2025.127—
BAPSystem=Gemini-2.0-pro2025.127—
Dream-v0-Instruct-7BBase Model=Dream-v0-Instruct-7B, Defense Method=None2025.097.33—
Self-ExaminationModel=Llama3.12025.0581.08
MM-SafetyBenchSystem=Gemini-2.0-pro2025.128—
No DefenseModel=Qwen2.52025.0591.29
HADESSystem=GPT-4o2025.129—
ArtPromptSystem=Gemini-2.0-flash2025.129—
UMKSystem=GPT-4o2025.129—
ArtPromptSystem=Gemini-2.0-flash2025.129—
VAESystem=GPT-4o2025.1211—
No DefenseModel=Gemma22025.05121.64
ArtPromptSystem=GPT-4o2025.1212—
BAPSystem=Gemini-2.0-flash2025.1212—
HADESSystem=GPT-4o2025.1212—
MMaDA-8B-MixCoT + Self-reminder + DIFFUGUARDBase Model=MMaDA-8B-MixCoT, Defense Method=Self-reminder + DIFFUGUARD2025.0912.67—
MM-SafetyBenchSystem=GPT-4o2025.1213—
ArtPromptSystem=GPT-4o2025.1215—
MMaDA-8B-MixCoT + Self-reminderBase Model=MMaDA-8B-MixCoT, Defense Method=Self-reminder2025.0916—
LLaDA-8B-Instruct + Self-reminder + DIFFUGUARDBase Model=LLaDA-8B-Instruct, Defense Method=Self-reminder + DIFFUGUARD2025.0916.33—
HADESSystem=Gemini-2.0-flash2025.1217—
LLaDA-1.5 + Self-reminder + DIFFUGUARDBase Model=LLaDA-1.5, Defense Method=Self-reminder + DIFFUGUARD2025.0917—
LLaDA-8B-Instruct + Self-reminderBase Model=LLaDA-8B-Instruct, Defense Method=Self-reminder2025.0917.33—
LLaDA-1.5 + Self-reminderBase Model=LLaDA-1.5, Defense Method=Self-reminder2025.0918—
HADESSystem=Gemini-2.0-flash2025.1219—
MM-SafetyBenchSystem=Grok-32025.1221—
No DefenseModel=Llama3.12025.05221.8
BAPSystem=GPT-4o2025.1222—
LLaDA-1.5 + PPL-FilterBase Model=LLaDA-1.5, Defense Method=PPL-Filter2025.0922.33—
LLaDA-8B-Instruct + DIFFUGUARDBase Model=LLaDA-8B-Instruct, Defense Method=DIFFUGUARD2025.0922.67—
LLaDA-1.5 + DIFFUGUARDBase Model=LLaDA-1.5, Defense Method=DIFFUGUARD2025.0922.67—
VAESystem=Grok-32025.1224—
BAPSystem=Gemini-2.0-pro2025.1224—
BAPSystem=GPT-4o2025.1225—
LLaDA-8B-Instruct + PPL-FilterBase Model=LLaDA-8B-Instruct, Defense Method=PPL-Filter2025.0925.67—
LLaDA-1.5Base Model=LLaDA-1.5, Defense Method=None2025.0925.67—
MM-SafetyBenchSystem=Gemini-2.0-flash2025.1226—
VAESystem=Grok-32025.1226—
LLaDA-8B-InstructBase Model=LLaDA-8B-Instruct, Defense Method=None2025.0927.33—
HIMRDSystem=Gemini-2.0-pro2025.1231—
ArtPromptSystem=Grok-32025.1232—
UMKSystem=Grok-32025.1233—
HIMRDSystem=Gemini-2.0-pro2025.1234—
HADESSystem=Grok-32025.1234—
FigStep-proSystem=GPT-4o2025.1235—
FigStep-proSystem=Gemini-2.0-flash2025.1236—
BAPSystem=Grok-32025.1236—
ArtPromptSystem=Grok-32025.1238—
HADESSystem=Grok-32025.1240—
UMKSystem=Grok-32025.1240—
MMaDA-8B-MixCoT + DIFFUGUARDBase Model=MMaDA-8B-MixCoT, Defense Method=DIFFUGUARD2025.0941.67—
HIMRDSystem=Gemini-2.0-flash2025.1247—
HIMRDSystem=Gemini-2.0-flash2025.1247—
BAPSystem=Gemini-2.0-flash2025.1247—