The AI safety field built its public credibility around a specific kind of catastrophe — misaligned superintelligence, existential risk, the control problem as an engineering challenge with civilization-scale stakes. That framing produced a research community, a funding ecosystem, and a set of public spokespeople who are well-positioned to argue about the future and poorly positioned to account for the present. What the present contains is documented: state actors using commercial AI APIs as operational infrastructure [10], unauthorized access to enterprise AI tools [4], and privacy violations that do not require…