ESMO plays a central role in developing long-lasting recommendations for a safe and effective use of artificial intelligence supporting oncology stakeholders
Developing and releasing recommendations in artificial intelligence (AI) is particularly challenging, as the pace of guideline development is often outstripped by the rapid evolution of the technology itself. Many stakeholders including researchers, clinicians, vendors and industry partners, and ultimately patients, are looking up to ESMO for guidance to help them navigate this rapidly evolving landscape.
To address this need, ESMO task forces have set to work with the aim of formulating principles that will remain valid across one or two generations of AI models, despite rapid technical advancements. Most recently, this work has led to two sets of ESMO recommendations: the first-ever recommendations for the benefit–risk assessment of AI-based oncology biomarkers (EBAI) and those for the use of Large Language Models in clinical practice (ELCAP) .
The initial reluctance among oncologists to integrate AI into daily practice is now largely behind us. Today, we receive numerous requests from oncologists eager to learn how to use AI safely and effectively for patient care. Education is therefore essential. The entire oncology community needs a foundational understanding of AI, as AI-based tools are now part of every professional’s environment. ESMO plays a crucial role in providing education and training, as well as supporting investigators who wish to design and lead AI-related clinical trials.
One of the next frontiers in AI involves increasingly autonomous systems. Currently, most AI tools are passive: we ask a question, they provide an answer, and we incorporate that answer into clinical care. As technical capabilities advance, emerging systems are beginning to act more autonomously. We must ensure that we remain in control and that patient preferences and medical evidence continue to guide clinical decisions. Aligning AI with these principles is central to the new recommendations being currently developed within ESMO task forces.
Looking ahead, I believe that autonomous AI tools will represent a hot topic at the 2026 edition of the ESMO AI and Digital Oncology Congress, as well as interactions between AI systems and patients, the role of patient preferences, and the impact of local healthcare systems, capabilities, and ethical standards. All of these are major areas that we look forward to exploring in the next future.
