Deep learning systems can be fooled by small, worst-case perturbations of their inputs, known as adversarial examples. This has been almost exclusively studied in supervised learning, on vision tasks. However, adversarial examples in counterfactual modelling, which sits outside the traditional supervised scenario, is an overlooked challenge. We introduce the concept of adversarial patients, in the context of counterfactual models for clinical trials—this turns out to introduce several new dimensions to the literature. We describe how there exist multiple types of adversarial example—and demonstrate different consequences, e.g. ethical, when they arise. The study of adversarial examples in this area is rich in challenges for accountability and trustworthiness in ML–we highlight future directions that may be of interest to the community.