How to Regulate AI?

After a string of surprising autonomous exploits by AI agents, calls for AI regulation are coming from all sides: industry, governments, consumers, and citizens. Some regions already have AI-specific laws, and AI is also subject to many existing laws that were never written with it in mind. But basic questions remain open. What should AI regulation aim to achieve? What exactly should be regulated, in what form, and who should be in charge? I will review current proposals and offer some answers to these central questions.