Lower tier AI risks such as deep fakes require defensive technologies and institutional adaptation. Open source use may limit defensive AI diffusion. Restricting widely available capabilities in a free society may do more harm than good. Technological innovation is not simply about acceleration or deceleration but involves branching paths. Survival during the AI transition will depend on the paths taken. Building AGI must be approached as a serious scientific endeavor. Epistemic rationality can be undermined by tribal affiliations. Different rationalist ideologies, such as EACC and EA, share a common goal of rationalism.