Updated Policy on the Use of Artificial Intelligence for the Journal of Neurosurgical Anesthesiology

The use of artificial intelligence (AI) in research and academic writing has dramatically increased in recent years, particularly since the introduction of generative AI tools such as ChatGPT (OpenAI), Gemini (Google) and Copilot (Microsoft). These tools represent a disruptive innovation in technology with important implications for academic publishing and research – good, bad, and ugly. The benefits are enormous, as these powerful tools can improve scientific writing and reduce language barriers for those writing in a secondary language. There are additional benefits of AI, such as predictive analytics using machine learning and large learning models (LLMs), data synthesis and knowledge translation that represent important progress in our knowledge base, ability to analyze and draw inferences, and ability to disseminate new evidence. However, the risks and boundaries on AI use must also be carefully recognized, and the integrity of the research and publishing process safeguarded. In this editorial, we present the updated policies and procedures for the use of AI for the Journal of Neurosurgical Anesthesiology. The Journal’s policy on the use of AI for authors has now been expanded and revised to reflect the current context and increased adoption of AI in academic work. These changes were informed by the Journal’s editorial board, a review of relevant literature1,2 and the current International Committee of Medical Journal Editors (ICJME) recommendations.3 The Journal permits the use of AI tools in the writing of the manuscript and the conduct of research; however this use must be transparently disclosed in the electronic submission process, title page, and methods section. Authors will now be required to provide a statement about the use (or non-use) of AI, including which AI tool(s) were use, and how they were used. If AI was used, authors must also state that they have reviewed the content and take responsibility for the results. This includes the use of AI in writing, data collection, data analysis (e.g. machine learning) and in the creation of figures, tables, or images. Manuscripts involving clinical prediction models using AI should follow the Transparent Reporting of a multivariable prediction model for Individual Prognosis Or Diagnosis (TRIPOD)+AI reporting guidelines guidelines4 (available at: https://www.tripod-statement.org/). Notably, in accordance with the ICJME policy, AI tools do not meet the criteria for authorship and cannot be listed as authors. Failure to disclose AI use may result in manuscript rejection due to lack of transparency, and authors must take full accountability for AI-generated material and results, and are responsible for fabricated, plagiarized or hallucinated content. Authors should refer to the Instructions for Authors available on the JNA website for the most current policy on AI when preparing their submission. The Journal’s policy also addresses the use of AI in the peer review process and aligns with other medical journals.5 Confidentiality is a critical the integrity of peer review, and manuscripts submitted to the Journal are considered privileged. Currently, the uploading of submitted manuscripts into AI tools or software is not permitted, as confidentiality cannot be guaranteed. However, reviewers may use AI technology to assist in peer review, such as to summarize evidence on a topic or identify similar articles, provided confidentiality is assured and critical assessment is provided by the reviewer. We acknowledge that AI and LLM tools may evolve in the future to leverage the benefits of AI and address many of these challenges, although the involvement of expert human peer reviewers will aways be required. Despite the growing integration of AI tools into biomedical research, several limitations warrant human oversight and vigilance. First, AI is susceptible to hallucinations, generating content that is plausible but factually incorrect, incorrectly referenced, or unsupported by the evidence.6 Such inaccuracies if unrecognized can misinform readers, distort evidence interpretation, and undermine research integrity. AI-generated text may convey an overly authoritative tone, even when the underlying evidence is weak, contradictory or absent. This may obscure uncertainty, oversimplify findings or mask methodological flaws, features critical to responsible conduct of research. Third, AI systems may reflect and magnify algorithmic biases in their training data, including biases related to geography, race, sex, socioeconomic status, disease prevalence or health-system structure.7,8 Such biases are particularly concerning in clinical research, guideline development and global health. Mitigation strategies include explicit bias checking by the authors, careful validation against primary sources, and a deliberate attempt to eradicate biases from medical science by promoting health equity.9,10 Overall, Artificial intelligence offers powerful opportunities to enhance efficiency, organization and clarity in scientific research and medical writing. However, these benefits are inseparable from their limitations and medical journals should proceed cautiously. The responsible integration of AI into scholarly work demands transparency, critical oversight, and clear boundaries. Maintaining scientific integrity requires the authors to remain fully accountable for the accuracy, and ethical implications of their work. By emphasizing disclosure, validation and human stewardship, the scientific community can harness AI’s potential while preserving the foundational principles of rigor, equity and trust upon which academic medicine depends.