Annette Flanagin

In September 2023, we announced the 10th International Congress on Peer Review and Scientific Publication, to be held in Chicago, Illinois, on September

As we have previously acknowledged, terminology, usage, and word choice are critically important in medical and scientific publication, particularly when

The Special Communication “Causal Inferences About the Effects of Interventions From Observational Studies in Medical Journals,” published in this issue

Reports on the use of artificial intelligence (AI) and machine learning, including large language models (LLM), in medical research have intensified in the last

Following the rapidly evolving use of artificial intelligence (AI) tools such as large language models (LLMs) and generative chatbots, JAMA and the JAMA Network

In Reply We thank Mr Uzoigwe and Dr Matsui and colleagues for their thoughtful comments in response to our Editorial.¹ Uzoigwe’s concerns about copyright of content generated by LLMs, neural networks, and similar AI tools have been widely discussed,^2,3 are in active litigation,³ and have been considered by the US Copyright Office⁴ and the World Intellectual Property Organization.⁵ Most agencies and jurisdictions require a human author to create or substantially contribute to the creation of the content for copyright to apply,^4,5 but there are ongoing discussions and often heated debates about the future of copyright and AI-generated content that go beyond legal interests, including concerns about ethics, plagiarism, and the displacement of jobs and elimination of creative professions. Companies that produce these tools have guidance regarding the reuse of generated content, but these may not be well understood by users. In addition, there is confusion about what defines “derivative” content and how “fair use” (which allows reuse of a small amount of text or graphic material for criticism, comment, news reporting, teaching, scholarship, or research)⁶ may or may not apply in the context of AI-generated content. To date, the US Copyright Office has denied copyright application for AI-generated work and also determined that a human-authored graphic novel combined with AI-generated images was copyrightable, but that the individual images could not be protected by copyright.⁴ The office has “launched an agency-wide initiative to delve into a wide range of these issues,” including public hearings.⁴ Hence, there is much uncertainty regarding rights to reuse AI-generated content in scholarly publication. We advise authors to respect the rights and licensing terms of any original content, to check the terms and conditions for reuse of any generated content, to be completely transparent about the use of any generated content (ie, explaining what content was generated and the names and versions of the tools used), and to be fully accountable for the content, including the rights to include such content in scientific publication.

In Reply We thank Dr Boffa and colleagues for their insightful comments in response to our Editorial describing expanded opportunities for researchers to share data in compliance with funder mandates.¹ We agree with their concerns about the rights of research participants and patients to privacy and risks of reidentification from data sets considered anonymous.^2,3 JAMA and the JAMA Network journals offer investigators a set of defined questions regarding their plans to share data when submitting manuscripts reporting research, which include whether the data to be shared are “deidentified participant data” or “participant data with identifiers.” While journals are responsible for ensuring that research participants and patients are not identified in published articles,⁴ journals are not responsible for the actual sharing of data or managing the risk of reidentification of individuals in shared anonymized data sets. Researchers, funders, and data providers must understand the risks of reidentification for data determined to be “deidentified”^2,3 and should consider the safeguard approaches recommended by Boffa and colleagues for appropriate data infrastructure and protocols to share data securely as well as proposals for researchers to assess reidentification risk⁵ and to manage and minimize that risk.⁶

Previous research has found limited progress in reporting race and socioeconomic status in studies published in medical journals from 2015 to 2019.¹ This study compared race and ethnicity reporting in 3 medical journals before and after implementation of Updated Guidance on the Reporting of Race and Ethnicity in Medical and Science Journals in August 2021.²

Artificial intelligence (AI) technologies to help authors improve the preparation and quality of their manuscripts and published articles are rapidly increasing