When you create an AI model to exploit vulnerabilities in an environment, guess what happens? It finds vulnerabilities in an environment. The issue seems to be that Anthropic didn’t think through all the implications of their testing and things got away from them. In other words, it was human error that was the proximate cause... Continue Reading