Anthropic Withholds New AI Model Over Cybersecurity Risks

Anthropic has caused a stir in the tech world by refusing to publicly release its latest AI model, citing real security dangers. Reports indicate the system can independently construct sophisticated digital attacks, not just advise on them but execute the dangerous steps itself. This development intensifies the debate on AI safety, putting pressure on governments, security agencies, businesses, and developers to enforce stricter access controls, clear safety regulations, and pre-release testing. The outlook remains tense: by 2026, the key question may no longer be about AI performance but about which models should be allowed to exist.