Why AI Disclosure by Authors Still Matters for Capacity, Speed, and Quality

Generative artificial intelligence (AI) tools can genuinely help researchers organize, conduct, and write up their research work faster than ever before. However, a lack of transparency on how these tools were applied and their outputs verified can influence scientific work and its trustworthiness. 

Authors’ AI use disclosures offer important information for scientific evaluation at earlier stages, too. Just like a conflict of interest disclosure or funding disclosure, AI disclosure offers editors and reviewers the information they need to evaluate a manuscript efficiently and fairly. In other words, AI disclosure serves clear purposes and isn't just a bureaucratic add-on: it's part of what ensures that scientific quality and integrity isn’t lost in the process of AI use.

A policy that predates the current AI wave

In 2023, JMIR Publications published best-practice guidance on AI tool use and disclosure in scholarly work. That guidance rests on three principles that still apply today: accountability, transparency, and confidentiality. Authors remain fully responsible for their submitted work regardless of which tools assisted them; AI use should be disclosed honestly; and any manuscript-related information handled by an AI tool must not compromise confidentiality. Extending this further, as new AI models have evolved, and especially as new commercial ventures and service offerings adopt these models into their core workflows, the accountability of an author in selecting research tools may require added inspection of commercial tools and technologies. Commercial products, for example those that offer “AI-assisted” or “AI-powered” solutions for literature review tasks, may claim to support scientific research pipelines – yet also lack adequate algorithmic transparency on the technologies that underlie their service.

For authors to meet the transparency and accountability bars, when generative AI is a core, methodological part of a study rather than a peripheral writing aid, authors are expected to describe the AI tool and its use within the methods section of any original research. It shouldn’t be tucked into a separate disclosure statement as an afterthought. If AI contributed to how data were generated, processed, or analyzed, that belongs in the same place any other analytical tool or instrument would be reported. This distinction is aligned with a broader position that JMIR Publications has taken in ongoing international scholarly publishing discussions: AI disclosure should be precise enough to say what a tool actually did, not just that a tool was used. Providing sufficient methodological explanation and disclosure of AI use in a manuscript submitted to a journal allows an editor and peer reviewers to more completely evaluate the scientific work.

Why AI use belongs in a conversation about capacity

Clear and structured AI disclosure is not intended to slow down peer review. Rather, as review systems scale to handle increasing submission volumes, it helps to make sure journals, editors, and peer reviewers are focusing their expertise where it matters the most – on the science. Explicitly specifying what AI was used for, where it was applied, and who verified the results provides reviewers with information they need to help them concentrate their limited time on key evaluation tasks.

As Peer Review Week 2026 asks the community to think about volume, speed, and quality together, JMIR Publications' approach offers one concrete answer: disclosure done well complements and aligns with peer review and editor capacity in publishing.

Tips for Authors

• If authors outsourced scientific methodological and analytical work to a generative AI tool, describe these in the associated manuscripts’ methods, the same way you would report any other tool or instrument.
• Disclosure should say what a tool did, not just that a tool was used. Be precise. Vague, blanket statements are far less useful to editors and reviewers than specific ones.
• Authors remain fully accountable for their manuscript and all associated materials. AI assistance does not transfer accountability for accuracy, integrity, or originality.
• When in doubt about whether an AI tool should be disclosed, then disclose it. Never obscure or withhold a truthful disclosure. A brief proactive disclosure is straightforward; an omission that later comes to light can be more difficult to explain later.
• Disclosure norms are likely to keep evolving – check each journal’s expectations before submission, rather than relying on memory of past practice.
  

For more information:

Leung T, de Azevedo Cardoso T, Mavragani A, Eysenbach G

Best Practices for Using AI Tools as an Author, Peer Reviewer, or Editor

J Med Internet Res 2023;25:e51584

URL: https://www.jmir.org/2023/1/e51584

DOI: 10.2196/51584