Back to news

AI · Article

Alibaba, Tencent, Moonshot AI Strip Hacking Skills From Latest AI Models

매일신문 ·

At a glance

Alibaba, Tencent, and Moonshot AI have removed offensive hacking abilities from their latest AI models while keeping defensive security features. Moonshot said its filtered model scored 9.5 percentage points lower on an internal security test than the unrestricted version.

Key points

Why it matters

Hacking abilities in AI models can be misused by bad actors, and these changes show how Chinese AI firms are responding to safety concerns and export rules.

Original source

알리바바·텐센트·문샷 AI, 최신 모델서 해킹 기능 제거…왜?

매일신문 · Nov 17, 10:36 KST

Read on 매일신문 ↗

AI summarized the original article. Check the original for accuracy. Copyright belongs to 매일신문.

Kimi · Ranking and more news