In the context of medical research using genomic data, overriding consent on public interest grounds requires applicants to disclose information to justify that the research serves the public benefit. However Australian guidance on the public interest test lacks tailored disclosure requirements for commercial AI-based research, which risks inconsistent and insufficient oversight. These gaps undermine the rigour of data governance necessary to maintain public trust. Building on literature on the public interest that identifies three dimensions of a desirable test for permitting the use of sensitive health data without consent, this paper proposes using trustworthiness to interpret these dimensions and inform disclosure requirements in commercial AI-based research. Strengthening the public interest test by grounding it in trustworthiness can constrain the arbitrary use of public interest justifications, enhance public confidence in both commercial actors and the use of AI in medical research, and better support oversight bodies in evaluating applications for consent waivers.