Researchers propose ‘radical optionality’ as a third way for AI regulation, where governments invest in tools for potential future crises, avoiding overregulation while building institutions and information channels. This approach prioritizes preserving democratic governments’ ability to make good decisions about AI governance. The key idea is to spend money and effort now to preserve options for an uncertain future. Why it matters: this approach could help governments navigate the immense stakes of AI development and mitigate potential risks, making it a crucial consideration in the broader AI landscape.