With the further development of artificial intelligence, it is becoming more complicated to draw a line between reality and digitally woven content. Deep-fake, the technology based on the executing of deep learning algorithms to create highly realistic manipulations of images, videos, and audio have attracted both its dangers and possible advantages. Although there is a wide opinion that deep-fakes have been used to facilitate misinformation, political manipulation and breach of privacy, there are also promising uses of deep-fakes in the medical, therapeutic and accessibility fields. Deep-fakes have the power to make the faces of loved ones visible to patients with Alzheimer or allow ALS patients to talk with their own voices as well as help patients in trauma or grief therapy with the help of a controlled virtual meeting. Applications of accessibility can be avatars of sign language, virtual socialization of the isolated, and speech restoration. The question this paper will answer is whether informed consent can be the adequate moral standard to support using deepfakes in the medical settings. It would also test the ethical justification of deep-fake therapy by analyzing it using the prism of two well-known theories on ethics Deontological theory suggested by Immanuel Kant and the Utilitarian theory suggested by Jeremy Bentham.