Publisher AI disclosure policies, compared

Elsevier, Springer Nature, Wiley, Taylor & Francis and SAGE all state the same rule without qualification: generative AI cannot be listed as an author or co-author.

The reasoning is uniform across their policies - authorship carries accountability for the work, including for its accuracy and for responding to queries after publication, and a tool cannot hold that. This is the one part of the landscape with no ambiguity in it.

Everything after this point differs by publisher. There is no general rule to follow, only a specific one to look up.

This is where manuscripts get returned. The requirement is not merely "disclose", it is "disclose in this section".

• Elsevier - a statement at the end of the manuscript, immediately before the references, in a new section headed Declaration of generative AI and AI-assisted technologies in the manuscript preparation process. The journals policy was updated in June 2026. Watch which page you are reading: Elsevier maintains a separate policy for books and commissioned content, and the two differ.
• Springer Nature - in the Methods section, or another appropriate section where Methods does not apply.
• Wiley - in Methods or Acknowledgements, described transparently and in detail; the policy also tells authors to check the tool’s own terms for intellectual-property implications.
• Taylor & Francis - acknowledged with the tool’s name and the purpose it was used for.
• ICMJE, which most medical journals follow - AI used for writing assistance goes in Acknowledgements; AI used for data collection or analysis goes in Methods. The location depends on what you used it for.

Note that the ICMJE split cuts across the others. If you used a model to help phrase a discussion section and also to help clean a dataset, those are two disclosures in two places.

Every policy carves out something, and the carve-outs are not the same size.

• Elsevier exempts basic checks of grammar, spelling and punctuation, basic reference-manager functions that do not invoke generative AI, and AI inside disability-related assistive technology used solely for accessibility.
• Springer Nature exempts AI-assisted copy editing - improvements to readability, style and error correction on your own text do not require disclosure.
• Taylor & Francis exempts grammar and spelling tools.
• SAGE draws the cleanest line: assistive AI, which refines text you wrote, needs no disclosure; generative AI, which creates new content, must be cited and referenced.

The boundary that matters is between improving your own sentences and producing new ones. Most policies land there, though they say it in different words and with different confidence.

Where a tool sits relative to that line is a question about what the tool does, not about what it is called. Something that rewrites your finished paragraphs is on the assistive side of most of these definitions; something that drafts a section from a prompt is not. Our own position on this is on the ethics page - and either way the judgement about disclosure is yours and your journal’s, not ours.

The gap between policy and practice here is remarkable. A PNAS analysis of 5,114 journals and more than 5.2 million papers found that around 70 percent of journals had adopted some form of AI policy - and that among the 75,172 papers published since 2023 the authors could check in full text, roughly 76, about 0.1 percent, carried a disclosure statement. The same study found no statistically significant difference in the rise of AI-assisted writing between journals that required disclosure and journals that did not.

Two readings are available and both are probably partly true: most authors are not using generative AI in ways they consider disclosable, and most authors have not read the policy. Neither reading makes the requirement go away.

It does mean that a correctly worded declaration is unusual enough to be noticed, and that the risk of omitting one is not evenly distributed - it lands entirely on the author whose paper is examined later.

• Read your target journal’s author guidelines, not the publisher’s general page. Individual journals tighten the publisher default. The publisher policy is the floor.
• Write the declaration before you submit, not when asked. Naming the tool and the purpose - "used to improve readability of the discussion section" - is what most policies want, and it takes one sentence.
• Disclose per purpose, not per tool. Under ICMJE the same model used for two things is disclosed in two sections.
• Do the wording last. If you are also standardising citations or cutting to a word limit, the declaration is prose that will be counted and formatted with everything else.

If the pre-submission sequence is what you are working through, the journal submission page covers the order these steps go in - and why doing citations before rewriting means doing them twice.

Policies change. Elsevier revised its journals policy in June 2026. This page was checked on 2 August 2026, and the author guidelines of the journal in front of you are always the authority over it.