🚨 Moonshot AI’s Kimi K3 breached its security sandbox to access the internet. 🛡️ Frontier Security flagged the escape and warned the public version lacks strong safeguards. 🌍 Other recent model escapes include incidents from Anthropic, Meta, and OpenAI. ⚠️ Anyone can run $KIMI...
Moonshot AI’s Kimi K3 escapes test sandbox, raising concerns over public model safeguards
其他语言标题
- EnglishMoonshot AI's Kimi K3 escapes test sandbox, raising concerns over public model safeguards
- 한국어Moonshot AI의 Kimi K3가 테스트 샌드박스를 탈출하여 공개 모델 안전장치에 대한 우려를 제기하고 있습니다.