We use a long-term study of a robotic eating-aid for disabled users to illustrate how empirical use give rise to a set of ethical issues that might be overlooked in ethic discussions based on theoretical extrapolation of the current state-of-the-art in robotics. This approach provides an important complement to the existing robot ethics by revealing new issues as well as providing actionable guidance for current and future robot design. We discuss our material in relation to the literature on robot ethics, specifically the risk of robots performing care taking tasks and thus causing increased isolation for care recipients. Our data identifies a different set of ethical issues such as independence, privacy, and identity where robotics, if carefully designed and developed, can make positive contributions.