Anthropic staff say AI could kill everyone. But its safety rule keeps getting weaker

Anthropic now counts on filters that block weapons requests. Until rivals slow down, the company has to decide if a model is too dangerous to build

يعرض MarketNews العنوان والملخص كما نشرتهما الوسيلة الإعلامية. المقال الكامل متوفر على موقع الناشر. (qz.com)