Beyond the Transparency Myth: Rethinking Information Duties for Medical AI
Seminar with Tatiana Aranovich, PhD researcher, Macquarie University, Australia.
Artificial intelligence is rapidly transforming healthcare, but how transparency should be designed, regulated, and used in practice remains deeply contested. While transparency is often presented as a cornerstone of safe and trustworthy medical AI, emerging research suggests that more information does not necessarily lead to better outcomes.
In this seminar, Tatiana Aranovich brings together comparative legal analysis and empirical insights to critically examine the role of transparency in medical AI governance. The presentation first explores how medical device frameworks in the EU, the United States, and Australia conceptualise and operationalise transparency, focusing on disclosures to regulators, healthcare professionals, and the public. It highlights comparative strengths across jurisdictions, including the EU’s data protection and AI governance approach, the United States’ regulation of adaptive AI and public databases, and Australia’s emphasis on clinical evidence, validation, and labelling obligations.
Building on this regulatory analysis, the seminar introduces the concept of a “medical AI transparency fallacy.” Drawing on qualitative interviews with healthcare professionals in Australia and New Zealand, it examines what information clinicians actually need—and use—in practice. While participants supported clear disclosure of AI use and key instructions for safe operation, the findings reveal significant barriers: clinicians may overlook labels, face limited AI literacy, and experience information overload.
The seminar argues that transparency alone is insufficient to ensure safe and accountable AI use. Instead, it proposes a holistic three-step framework: (1) improved information design tailored to clinical contexts, (2) stronger information supply through medical education and professional guidance, and (3) enhanced regulatory disclosures to enable external scrutiny and accountability.
By combining doctrinal and empirical perspectives, this seminar offers a timely and critical rethinking of transparency in medical AI. It will be of interest to researchers, regulators, clinicians, and industry actors seeking to design more effective and realistic governance frameworks. Participants will gain insight into both the promises and limits of transparency, and practical pathways for improving the safe implementation of AI in healthcare
Speaker bio
Tatiana Aranovich is a PhD researcher at Macquarie Law School, Macquarie University (Australia), specialising in the regulation of artificial intelligence in healthcare. In 2024, she interned at the Therapeutic Goods Administration (TGA), where she contributed to regulatory reform by producing a report on AI transparency and explainability in medical devices.
She teaches in the areas of Law and Technology, Contracts, and Torts, and works as a research assistant on the ARC-funded projects “No to Black Box: Toward Transparent and Safe AI in Healthcare” and “Generative AI and Creative Industries”. With over 16 years of professional experience—including eight years at Brazil’s national health regulator—she brings a multidisciplinary and practice-oriented perspective to questions of public sector innovation and digital health governance.