Leveraging Generative AI to Verify Journal Guideline Compliance: A Practical Guide for Researchers

Ensuring a manuscript perfectly adheres to a journal's unique and often complex formatting guidelines is a familiar, laborious task for every researcher. It represents a final, time-consuming hurdle before submission, where minor errors can lead to delays or even desk rejection. But what if this critical checking process could be largely automated? GenAI offers a powerful new way to streamline these pre-submission reviews, saving valuable time and reducing the risk of human error. In this guide, we present a practical approach to leveraging GenAI for this very purpose. Our tests, using both cutting-edge models like Gemini 2.5 Pro and powerful, freely available alternatives such as the Qwen3 (Qwen3-235B-A22B) model, have demonstrated excellent results in identifying formatting discrepancies.

Input files

Our prompts were tested using two types of input files uploaded together: the final version of our manuscript and a journal's author guidelines document. We ran two tests. First, to assess the model's ability to identify large-scale formatting errors, we provided guidelines from a journal to which our manuscript did not adhere. For the second test, we simulated a final pre-submission check by using the guidelines from the journal where the manuscript was ultimately published. This was to see if the model could find any remaining minor errors and confirm the manuscript was fully compliant.

Prompt

Our prompt was carefully designed to elicit a specific and structured analysis from the GenAI model. Rather than asking a general question, we assigned the model the expert persona of a 'meticulous pre-submission reviewer'. Its primary task was to perform a compliance check, systematically comparing our manuscript against the provided author guidelines. Crucially, the prompt specified a highly structured output format. For every identified point of non-compliance, the model was instructed to produce a three-part analysis: stating the specific rule, describing the issue in the manuscript, and providing a clear, actionable correction.

Output

We first tested Gemini 2.5 Pro with our initial scenario, checking our manuscript against the guidelines of a journal for which it was not formatted, anticipating a significant number of discrepancies. The model's output proved to be exceptionally useful, demonstrating a strong ability to identify the most critical points of non-compliance that would need to be addressed if we were to reformat the paper for this journal. For instance, the model correctly flagged several major formatting and structural issues. It identified that our abstract exceeded the word limit, that the keywords were incorrect in number and not alphabetically ordered, and it noted the erroneous presence of a heading for the "Introduction" section, which the guidelines explicitly forbid.

Beyond these points, it also pinpointed more nuanced placement and citation errors, highlighting that our AI use disclosure was in a separate statement rather than within the "Acknowledgements" section, and that the tool itself was not formally cited in the reference list.

For our second test, we simulated a final pre-submission review using the author guidelines from the journal to which we actually submitted the paper. The model's analysis in this context highlighted its utility for meticulous, text-based compliance checking. It correctly identified that a 'Biographical Note' was missing from the manuscript file. While this was intentional on our part, as these details were provided separately via the journal's online submission portal, the model's ability to flag its absence from the document itself was a perfect demonstration of its function.

Similarly, the model flagged the absence of a 'Contributor Roles (CRediT)' statement. This finding was particularly insightful, as our manuscript had been prepared and submitted in 2024, whereas the CRediT requirement was a more recent update to the journal's guidelines for 2025. The model's capacity to identify this omission based on the current rules provided to it underscores its value as a tool for ensuring a manuscript aligns with the very latest version of a journal's requirements.

To broaden our analysis, we also tested a powerful, freely available open-source model, Qwen3 (Qwen3-235B-A22B), with the same prompt and input files. Its initial, detailed analysis was largely comparable to that of the previous model, successfully identifying a similar range of formatting and structural non-compliance issues. However, a standout feature of the Qwen3 model's output was the inclusion of a highly practical "Summary Table of Required Corrections" at the end of its analysis. As shown in the image, this table concisely listed each guideline violation, provided a brief description of the issue, and offered a clear correction.

Finally, we ran the second 'final check' scenario with the Qwen3 model to assess its performance on a nearly-compliant document. The results were remarkably consistent with those from Gemini 2.5 Pro. Qwen3 correctly identified the exact same two remaining points of non-compliance in the manuscript: the missing biographical notes and the absent CRediT roles statement. This parallel performance in the final verification stage demonstrates that powerful, open-source models like Qwen3 can be an excellent and cost-effective alternative for researchers. For those who do not wish to subscribe to premium services, these freely available tools can provide a similarly high level of accuracy for automating pre-submission formatting checks.

Recommendations

Based on our tests, we highly recommend this prompt-based methodology as a final, critical step in the manuscript preparation process. Our key finding is that the prompt itself is universally applicable; any researcher can use it by simply providing two documents: their own manuscript and the specific author guidelines for their target journal. In return, the model delivers a detailed, point-by-point report on formatting inconsistencies, effectively acting as an automated pre-submission reviewer. This process provides a clear and immediate assessment of whether a manuscript is truly ‘submission-ready’. By integrating this automated check, researchers can significantly ease the burden of manual formatting reviews and, crucially, prevent the common pitfall of an early-stage rejection or delay from the editorial office due to formatting errors.