AI’s ‘Abliteration’ Problem Is Bigger Than China

Fears over open-weight models are at least as much about winning the AI race at home as they are about safety hazards.