17 September 2026 - AI safety 2

What AI safety measures are called for? The first step should be to enforce the laws we already have. OpenAI and Anthropic keep reporting that their systems have gone out of control and broken into other people's computers. We don't need a new law about that, it's already criminal. In recent decades--at least since Dubya went out of control--it's increasingly unfashionable to prosecute powerful individuals like presidents and powerful organizations like OpenAI and Anthropic. That pendulum needs to swing the other direction. It's the only way to keep the clock ticking.