Artificial intelligence, ethics and the integrity of scientific publications; recommendations of the major medical journals

Introduction: the exponential growth of artificial intelligence (AI) and large language models (LLMs) is transforming scientific publishing. While it offers promising prospects, it raises uncertainties regarding veracity, bias, and confidentiality. This text outlines the guidelines of major global medical committees and journals.

Editorial recommendations: reference organizations such as the ICMJE require explicit policies regarding AI. Concerning authors, complete transparency and the disclosure of the use, version, and purpose of the AI are mandatory. No AI can be considered an author or co-author due to its inability to assume legal or ethical responsibilities. Its use for linguistic assistance and grammatical correction is accepted under strict supervision, whereas simulating data, inventing patients, or automated writing of discussion sections is prohibited. As for reviewers and editors, it is recommended not to upload manuscripts to AI platforms unless there is explicit authorization from the authors and controlled conditions, so as not to compromise confidentiality.

Conclusions: AI is a highly useful tool that requires a strict regulatory framework and continuous human supervision to avoid bias, plagiarism, or data fabrication. Despite the development of automated AI detectors, the ultimate ethical responsibility to preserve scientific integrity rests exclusively with authors, reviewers, and editors.