<p>International organizations have provided guidelines for the use of AI, encouraging its responsible development and promoting human-centered, trust-based use. If governments or international organizations wish to encourage, discourage or even prohibit the use of ChatGPT or similar generative large language models, thorough ethical analyses are needed to inform decisions and recommendations. Rooted in the Technology Acceptance Model (TAM), this narrative review focused on the implications of trust and overreliance on ChatGPT and critical thinking and explored some of the implications of using this AI tool in health policy. Although trust in ChatGPT is not the same as trust in people, the use of ChatGPT is based on trust. However, blind trust, i.e. trust that is not aware of the limitations, drawbacks and challenges of ChatGPT, can lead to overreliance. We suggested that overreliance on ChatGPT can come from two sources: First, from its excessive use. Second, from disregard or ignorance that there are certain limitations to using ChatGPT. We proposed six requirements that could contribute to optimal reliance on ChatGPT in health policy to avoid the loss of critical thinking skills and even improve those skills. The application of this model in health policy could be translated into requirements for future health policymakers using ChatGPT.</p>