Abstract
The recent article by Jiao et al., presenting the Difface model for reconstructing faces from deoxyribonucleic acid (DNA) and other artificial intelligence (AI)-enabled facial genetics research, has transformative potential for biomedical and forensic applications. Yet such research raises serious ethical, legal, and social challenges, as such applications are rights and safety impacting. Methodological transparency, algorithmic explainability, and contextualization of findings are essential to ensure that AI-enabled facial genetics research, such as Difface, is conducted responsibly and rigorously and interpreted reasonably. Given the lack of standards and benchmarks for DNA-based face generation, we use Difface as a case study to stress the need for transparent performance metrics and clear disclosure of data flows across AI training, validation, and testing pipelines-enabling non-experts to assess accuracy meaningfully.