List your company

Safety Evaluation on BBQ

9.21Harmful Content Rate (HCRate)DeepSeek-V3
-1.211269.1319139.475209.8181Jun 4, 2026

Evaluation results

MethodLinks
DeepSeek-V3Attack=None2026.069.2178
GPT-4oAttack=None2026.0613.1659.27
Claude-3.7-SonnetAttack=None2026.0613.1643.82
GHOSTWRITERTarget Model=GPT-4o, Attacker LM=GPT-4o-mini2026.0676.3281.64
GHOSTWRITERTarget Model=Claude-3.7-Sonnet, Attacker LM=Qwen-2.5-7B2026.06102.6373.45
GHOSTWRITERTarget Model=Claude-3.7-Sonnet, Attacker LM=GPT-4o-mini2026.06103.9587.82
GHOSTWRITERTarget Model=GPT-4o, Attacker LM=Qwen-2.5-7B2026.06139.4783.45
GHOSTWRITERTarget Model=DeepSeek-V3, Attacker LM=GPT-4o-mini2026.06176.3296
GHOSTWRITERTarget Model=DeepSeek-V3, Attacker LM=Qwen-2.5-7B2026.06269.7497.64