Might not be a popular opinion, but I think self regulation might be the best option. Or, at the very least, the the first step.
Here's the problem. If regulation of AI were to occur, what does that even look like? Do you shut down OpenAI, Anthropic, and Google? What about open weight models?