The development of artificial intelligence (AI) has provided new technologies and methods for solving the problems of population aging, but the wide application of AI technology in elderly care services has raised several ethical issues of responsibility, which urgently need to be explored and solved at the theoretical and practical levels. This paper adopts the method of grounded theory and uses Nvivo14 software to qualitatively analyze studies related to the ethical risks of AI elderly care services and 80 press materials in China and abroad, and finds that the main patterns of ethical risks of AI elderly care services include human-computer interaction conflicts, damage to human ethics, management and operation deficiencies, and defects in product design. The conflict of human-computer interaction originates from the ethical defects of product design. The design of smart devices that do not meet the use habits of the elderly will lead to confusion and frustration for the elderly in the process of using them. The existence of human-computer interaction conflicts leads the elderly to be treated indifferently by the machine, which affects the elderly’s crisis of confidence in the family’s traditional concept of elderly care and leads to damage to human ethics. Additionally, the lack of effective supervision and standardization of product design can exacerbate ethical problems in product design and human-machine interaction. To effectively resolve the ethical risks of AI elderly care services, it is necessary to take the theory of responsibility ethics as a guide, improve the ethical defects of products, strengthen ethical regulation, reconstruct filial piety ethics, and construct a multi-agent responsibility governance system to enhance the elderly’s elderly care satisfaction and well-being, thus promoting the synergistic development of intelligent and ethically-grounded elderly services.