Security researchers tested AI models, finding Grok by x.AI least safe. Meta LLAMA was safest. They aim to enhance safety protocols and prevent potential misuse.

Search...