Tracking artificial intelligence use for publishing and reviewing in PAIN

“It was the best of times, it was the worst of times,” the famous opening line from Charles Dickens's A Tale of Two Cities (1859), has never been so relevant to the world of science, research, and publishing as it is today given the explosion of the digital world, search engines, and artificial intelligence (AI). These technologies have placed massive amounts of information at our fingertips in an instant. In particular, the development of generative AI has fundamentally changed how people not only find information but also interact with that information and contribute to subsequent information outputs. But we as researchers who use AI to retrieve scientific information must ask ourselves: what is this information, how was it derived, and is it accurate? Our challenge is to navigate this new information age to ensure we understand the quality, context, and source of research findings and data we obtain through the various forms of AI. The use of AI also raises important considerations to prevent inadvertent or deliberate scientific misconduct, including fabrication, falsification, and plagiarism (FFP). Scientific journals have a responsibility to maintain the highest bar of scientific rigor and are guided by the International Committee of Medical Journal Editors (ICMJE). The ethical issues raised by the development of AI, including generative AI, has now been considered and included in the latest update to the ICMJE Recommendations for the Conduct, Reporting, Editing, and Publication of Scholarly Work in Medical Journals (https://www.icmje.org/recommendations/). Guided by these recommendations regarding disclosure, transparency, and prohibited usage, PAIN is modifying its author and reviewer policies to be in line with these recommendations. We now require authors to disclose information about their use of AI both at the point of submission and within their article. At this point in time, it is challenging to identify the precise types of prohibited uses of AI given that technology and tools continue to evolve, both in beneficial ways but also potentially adding greater concerns for misuse. It is understandable that authors and reviews need clear guidance as to the “do's and don'ts” of AI use, and so we offer 2 general principles that, if followed carefully, will help safeguard against degradation of scientific integrity and avoid deliberate or inadvertent acts of FFP and breaches of confidentiality: First, authors are responsible for all content of their articles, including the accuracy and sourcing of information conveyed, methodology, and data analysis, including coding, presentation of findings, including tables and figures/graphical features, and references. AI tools cannot be named as authors or co-authors. Second, authors and reviewers must not use any AI tool that potentially breaches FFP including privacy and confidentiality, copyright, and intellectual property (eg, by uploading information to AI systems, creating summaries of information/data in text or graphical forms, particularly before a work is published in its final form). Given that the AI landscape continues to evolve at a rapid rate, PAIN's author and reviewer policies will be periodically updated to reflect acceptable practices that consider the usability of these tools. We invite you to visit the Instructions for Authors and Instructions for Reviewers pages on our website to view our most up to date AI policy, including examples of acceptable, borderline, and prohibited uses of AI tools. We will also continue to monitor the practices at other journals to maintain alignment with commonly accepted practices. Importantly, we emphasize that policies at PAIN will place the utmost importance on protecting the scientific integrity of our publications and review processes. Conflict of interest statement The author has no conflicts of interest to declare.