앤트로픽 “해킹 질문엔 답변 제한”…‘미토스’ 안전 모델 일반 공개

AI safety company Anthropic has announced that its new safety model, "Mitos," is now publicly available. Mitos is designed to prevent AI systems from generating harmful or inappropriate content, particularly in response to queries related to hacking or illegal activities. The model aims to enhance the responsible development and deployment of artificial intelligence. Anthropic stated that Mitos will limit responses to such questions, thereby mitigating potential misuse.

This development is significant as it introduces a new tool to proactively address safety concerns in AI, promoting more secure and ethical AI applications.