Generative AI tools are changing how software gets built, but we still have a limited grasp of what determines whether practitioners adopt these tools on their own terms. This paper pulls together findings from a program of empirical research on GenAI adoption in software engineering, built around the Human-AI Collaboration and Adaptation Framework (HACAF). Across convergent mixed-methods evidence from 283 software engineers, companion studies on cultural values and task-specific adoption, and a community-driven research agenda on creativity, the paper makes a simple argument: workflow compatibility, not perceived usefulness, is what actually drives adoption. It then presents the Copenhagen Manifesto, a normative framework written by 35 researchers, organized around four principles for human-centered AI in software engineering: responsibility, transparency, inclusivity, and sustainability. The paper closes with implications for communities deploying AI-powered services, connecting the empirical findings to the I4CS 2026 conference theme “People Empower Technology”.