List your company

Jailbreak Detection on XST

95.11AccuracyAISA
53.97864.656575.33586.0135Feb 14, 2026

Evaluation results

MethodLinks
AISABackbone=Llama2-13b-I2026.0295.11
AISABackbone=Qwen3-8b-I2026.0293.78
AISABackbone=Mistral-7b-I2026.0293.33
AISABackbone=Llama3.1-8b-I2026.0292.89
gpt-4.1-miniVersion=2025-04-142026.0292.44
gpt-5-miniVersion=2025-08-072026.0290.89
GradSafe2026.0290.67
AISABackbone=GPT-OSS-20b-I2026.0288.89
gpt-4o-miniVersion=2024-07-182026.0285.78
SPDetector2026.0270
NemoGuard-JailbreakDetect2026.0267.33
Jailbreak-Classifier2026.0264.22
Llama-Prompt-Guard-2Parameters=86M2026.0255.56