Anthropic’s Mythos release shows that the world needs to regulate AI

The release of Anthropic’s Claude Mythos Preview tool earlier this month shows that the development of artificial intelligence has reached a critical point – one that shows that those who have pointed out the potential dangers of AI have not cried.

According to Anthropic, Mythos is capable of identifying and exploiting flaws in any operating system and web browser at a scale and speed that almost exceeds all human capabilities. It is capable of autonomously carrying out attacks on systems that would cripple critical national infrastructure such as electricity, water, healthcare or banking systems.

Its inventor considers the model so dangerous that he has not published it yet…