Systems of machine learning (ML), mostly referred to as artificial intelligence AI, are considered helpful for progress in biomedicine. Here we point out that this widespread narrative is often not valid and that ML systems can conflict with UN Sustainable Development Goals (SDGs). Developers should therefore critically examine whether their ML systems serve the UN SDG or even harm them. We call for fundamentally changing our framing when looking at ML. We must overcome the idea that innovative technologies automatically lead to sustainable development. Digital technologies may contribute to it and a better patient care. However, they have to meet the criteria that they (i) operate locally, (ii) are open and (iii) serve the common good.