December 2025: The Legal and Ethical Framework for AI in Gastrointestinal Endoscopy: A WEO International Consensus Statement

The WEO has released an international consensus statement that sets out how artificial intelligence (AI) should be used responsibly in gastrointestinal (GI) endoscopy. Published in Annals of Internal Medicine, the statement focuses on three key areas: protecting data, clarifying medicolegal responsibilities, and ensuring fairness and equity in AI applications.

The recommendations were developed through a two-round Delphi process involving 14 experts from 11 countries. This work was carried out under the OperA project — Optimising Colorectal Cancer Prevention through Personalized Treatment with Artificial Intelligence — which is funded by the European Commission. The expert panel ultimately agreed on 10 statements.

• Data governance is a central theme. The group stresses that all AI systems used in GI endoscopy must follow local data-protection rules, keep patient information private through anonymization, and be open about any algorithm updates and performance results.
• In discussing medicolegal issues, the statement notes that clinicians are responsible for using AI tools in line with established clinical standards and manufacturer instructions. It also calls on professional societies to define clear liability guidelines covering AI-supported diagnosis, automated reporting systems, and new performance indicators that may emerge as AI is integrated into practice.
• The consensus also addresses equity and bias, urging that AI algorithms be validated using diverse patient data. It recommends transparent reporting of the demographic characteristics of training and testing datasets, along with research into whether the use of AI could inadvertently worsen existing health disparities.

Taken together, these consensus statements provide a groundwork for future guidelines and regulations aimed at ensuring that AI is adopted in GI endoscopy in a safe, ethical, and equitable way.

We would like to thank all the authors who contributed to this work: Omer F. Ahmad, Yuichi Mori, Michael Bretthauer, Daniel A. Dourado, Cesare Hassan, Raf Bisschops, Pradeep Bhandari, Michael F. Byrne, Evelien Dekker, Uma Mahadevan, Fola P. May, Helmut Messmann, Masashi Misawa, Haruhiko Ogata, Yutaka Saito, Anna L. Silverman, Pu Wang, Tomonori Yano, Lars Aabakken, and Tyler M. Berzin.

The document is available here.