Startups such as Hirundo are editing Chinese open-weight models to strip censorship for Western businesses

Editorial illustration published by Rest of World with its story on realigning Chinese AI models — Rest of World illustration

Rest of World reports that firms including Hirundo and Lazarus use machine unlearning and direct weight edits on models from DeepSeek, Moonshot, Z.AI, Alibaba and Tencent to remove state-aligned bias and harden security.

Read at Rest of World

Sources

Rest of World

via @Techmeme

Editor: Haydn Snape

More on AI safety

Eight independent AI evaluation groups form a forum to set ground rules for testing frontier models
WIRED finds about 400 apps posing as ordinary AI tools that hide sexually abusive deepfake generators
TÜV groups propose independent pre-launch testing of frontier AI and agents, led by Germany's AI safety institute
Microsoft AI chief Mustafa Suleyman says labs must not build conscious superintelligence
VicOne reports 28 zero-day findings in robotics AI software in under two days using AI-assisted research