AUSTRIA Law and Practice Contributed by: Harald Strahberger and Florian Sesztak, Kinstellar
9.4 Emerging Legal Challenges As healthcare AI technologies advance, Austria is fac - ing several emerging legal and regulatory challenges. One key issue is the integration of adaptive or continu - ous learning systems, which conflict with the MDR’s requirement for fixed intended performance. Austrian regulators are still determining how to monitor and recertify algorithms that evolve post-deployment. Another challenge involves autonomous AI systems, especially where AI plays a central role in diagnosis or treatment planning. The AI Act will require robust human oversight mechanisms, but enforcement strat - egies are still being developed. Questions around traceability, explainability, and accountability remain open, open-particularly for “black box” algorithms used in clinical contexts. In addition, since the ÄrzteG reserves core diagnos - tic and therapeutic decisions to licensed physicians, AI systems may only assist, but not replace, medical decision-making in clinical settings. Any attempt to delegate or automate such functions without mean - ingful human involvement could therefore violate national professional and liability laws. Finally, the growing convergence of AI with robotics, augmented and virtual reality, and digital therapeu - tics raises issues around joint regulatory classification, dual certification, and cross-sector liability. Austrian regulators are preparing by participating in EU expert groups, drafting national implementation plans for the AI Act, and encouraging interdisciplinary research into AI governance. As a result, developers are being advised to incorporate legal, ethical, and technical safeguards into their systems now to prepare for future scrutiny and potential legal risks.
data protection impact assessments, and maintain - ing comprehensive documentation covering algorithm functionality, updates, and risk assessments. The AI Act will require risk management, data governance, human oversight design, and transparency documen - tation. Organisations should establish multidiscipli - nary AI compliance committees comprising experts from legal, clinical, IT, and data protection fields. Doc - umentation must cover the training data used, bias mitigation efforts, system logic, and update controls. To balance innovation and compliance, institutions should engage in regulatory sandboxes (eg, via FFG or Horizon Europe), use phased clinical rollouts and consult early with BASG. 10.2 Contracting and Liability Allocation Contracts must now anticipate obligations under the AI Act, in addition to the MDR, GDPR and the national legislation. Key provisions include a clear allocation of regulatory compliance, as follows. Developers are responsible for AI-specific risk man - agement, data training quality, and conformity docu - mentation, while healthcare institutions manage the clinical use and integration of AI. Indemnities must reflect this, covering algorithmic malfunction, data breaches, or non-compliance penalties. Limitation of liability clauses remain common but may be adjusted upward due to potential fines under the AI Act (up to 6% of global turnover). Contracts should specify responsibilities for post-market monitoring, incident reporting, and software updates, which are mandatory under both MDR and the AI Act. Where adaptive or continuously learning systems are used, provisions must define who ensures re-certification and when. As transparency and explainability will be legally required, warranties may need to guarantee the availability of documentation for clinicians and patients. 10.3 Insurance Considerations Healthcare AI stakeholders in Austria should consider a combination of professional indemnity, product lia - bility, and cybersecurity insurance. These cover clini -
10. Practical Considerations in Healthcare AI 10.1 Compliance Strategies
Healthcare AI developers in Austria must establish robust compliance structures that align with the MDR, the GDPR, the AI Act, and Austria’s national health - care laws. Recommended strategies include imple - menting a quality management system, conducting
29
CHAMBERS.COM
Powered by FlippingBook