Dario can only see the danger of open weights models, and refuses to comprehend the real and actual danger of a concentration of power for those who deem themselves philosopher kings to control who has access to intelligence, how much intelligence, and when. "Open-weights models that don’t have dangerous capabilities are a public good" This man would advocate to ban programming and the internet if he was born a little earlier. And there's still time in his life to do it later, especially if he is allowed to control intelligence. We've already seen just how much he has disdain for public squares. We desperately need to not let this guy get his way. "Open-weights models—it does not matter whether they come from China or anywhere else—do potentially present a higher risk than closed models, because it is very difficult to apply guardrails to them or monitor their usage, and once weights are released they cannot be withdrawn" Notice his entire viewpoint hinges on some person, or body, deciding who should have access to intelligence. The issue here is thinking some small group of people is somehow the safer path. Who are these people? why are they safer/better/smarter? This concentration of power is the actual dystopic danger. All of the problems we actually are seeing in AI right now are the result of closed models getting abused, not being useful for defense, and all guardrails being circumvented at scale. "All sufficiently capable models, open and closed, should go through mandatory safety testing." We have already seen how this worked out with OpenAI. We cannot and should not be reliant on one or two companies somehow being "safe" for us. We are already testing out this theory. It's going terribly. Anthropic has never released a model that people couldn't jailbreak. They speak about distillation "attacks" but this is a double edged sword. Clearly, allegedly, Anthropic is self-reporting as being incapable of stopping widespread guardrail defeat at immense scale.