The European Society for Medical Oncology (ESMO) has developed the ESMO Guidance on the Use of Large Language Models in Clinical Practice (ELCAP), released during the ESMO Congress 2025 in Berlin, to establish structured standards for the safe and effective integration of AI language models in oncology.

ELCAP was created through a collaborative, interdisciplinary process involving experts in oncology, AI, ethics, and patient advocacy from November 2024 to February 2025, addressing the complexities of AI adoption in healthcare.

The guidance categorizes AI applications into three types: patient-facing tools like chatbots, clinician decision support systems requiring validation and transparency, and institutional background systems linked to electronic health records, each with specific validation and oversight protocols.

ELCAP emphasizes the importance of context-specific AI use, highlighting supervision, data quality, transparency, and clear escalation pathways to ensure AI outputs are reliable and support clinical decision-making.

The framework underscores that AI tools are meant to assist clinicians, not replace them, and stresses maintaining trust in clinical judgment while adopting these technologies.

It also highlights that the quality of AI output depends on input data quality and that supervision, governance, and clear escalation pathways are essential to prevent errors and biases.

The guidance discusses future challenges posed by autonomous AI models that can act independently, raising safety, regulatory, and ethical issues, and calls for shared standards to ensure responsible AI use in oncology.

ESMO President Fabrice André emphasized that shared standards are crucial alongside algorithms to maintain trust and improve the quality, equity, and efficiency of cancer care through AI.

ELCAP advocates for assistive large language models operating under human oversight to support clinical decisions, while acknowledging the emerging risks of autonomous AI models that require additional regulatory guidance.

This initiative aligns with the broader focus of the ESMO Congress 2025 on AI’s transformative role in cancer care, aiming to enhance treatment quality, equity, and efficiency while safeguarding patient welfare.