Recommendation 1: Chatbots cannot be authors. Chatbots do not meet the criteria for authorship, including the ability to give final approval of the version to be published and to be responsible for all aspects of the work in ensuring proper research and resolving issues related to the accuracy or integrity of any part of the work. No artificial intelligence tool can “understand” a conflict of interest statement and does not have the legal standing to sign such a statement. Chatbots have no affiliation with an organization independent of their developers. Since authors submitting a manuscript must ensure that all authors meet the criteria for authorship, chatbots cannot be included as authors.
Recommendation 2: Authors should be honest about their use of chatbots and provide information about how they were used. The extent and type of use of chatbots in scientific publications should be reported. This is consistent with the recommendation to acknowledge assistance in writing the article and provide detailed information in the article about how the research was conducted and how the results were obtained.
Recommendation 2.1: Authors submitting an article in which a chatbot/AI was used to develop new text should acknowledge such use in the acknowledgements section; any prompts used to generate new text or to convert text or text prompts into tables or illustrations should be reported.
Recommendation 2.2: If an AI tool such as a chatbot was used to perform or generate analytical work, assist in reporting results (e.g., creating tables or figures), or write computer code, this should be reported both in the abstract and in the body of the article. In the interests of enabling scientific control, including replication and detection of falsifications, the full prompt (query operator) used to generate the research results, the time and date of the query, and the AI tool used and its version should be provided. Authors are responsible for the material provided by the chatbot in their article (in particular, for the accuracy of the material presented and the absence of plagiarism) and for appropriately citing all sources (including the original sources for material generated by the chatbot). Authors are responsible for ensuring that the content of the article reflects the authors' data and ideas and is not plagiarized, fabricated, or falsified. Otherwise, it is a violation of scientific norms to propose such material for publication, regardless of how it was written.
Recommendation 3: Authors are responsible for the material provided by the chatbot in their article (including the accuracy of the material presented and the absence of plagiarism) and for properly citing all sources (including the original sources for material generated by the chatbot). Authors are responsible for ensuring that the content of the article reflects the authors' data and ideas and is not plagiarized, fabricated, or falsified. Otherwise, it is a violation of scientific norms to propose such material for publication, regardless of how it was written. Similarly, authors should ensure that all cited material is properly cited, including full citations, and that the cited sources corroborate the text produced by the chatbot. Since the chatbot may be designed to avoid using sources that contradict the views expressed in its output, authors are responsible for finding, reviewing, and incorporating such opposing views into their articles. Authors should indicate what they have done to reduce the risk of plagiarism, provide a balanced perspective, and ensure the accuracy of all their references.
Recommendation 4: Editors and reviewers should disclose, both to authors and to each other, any use of chatbots for manuscript evaluation and review and correspondence. Editors and reviewers are responsible for any content and citations produced by the chatbot. They should be aware that chatbots store the prompts sent to them, including the content of the manuscript, and providing the author’s manuscript to a chatbot violates the confidentiality of the manuscript submitted for publication.
Recommendation 5: Editors should, where possible, use appropriate tools to help them detect content created or modified by AI. Such tools should, where possible, be used by editors for the benefit of science and the public.