kakaocorp/kanana-safeguard-siren-8b
TEXT GENERATIONConcurrency Cost:1Model Size:8BQuant:FP8Ctx Length:8kPublished:May 26, 2025License:apache-2.0Architecture:Transformer0.0K Open Weights Cold

The kakaocorp/kanana-safeguard-siren-8b is an 8 billion parameter language model, based on Kakao's Kanana 8B, specifically designed for detecting legal and policy risks in conversational AI systems. It classifies user utterances into predefined risk categories (Adult Content, Professional Advice, Personal Information, Intellectual Property) and outputs a single token indicating SAFE or UNSAFE with a risk code. Optimized for Korean, this model excels at identifying potentially problematic user inputs to enhance AI safety.

Loading preview...