Powerful AI should not be released without safeguards, National Cyber Security Centre says

It follows Anthropic's decision not to release Mythos, which is capable of identifying and exploiting weaknesses across 'every major operating system and every major web browser', on a general basis