Can AI Be Listed as an Author on a Paper?

In January 2023, at least four medical journal articles listed ChatGPT in the author byline, credited alongside the human researchers who wrote the paper. Editors caught it within weeks, and every one of those credits was removed. Can AI be an author? The episode already answered the question once. What it does not explain is why the answer is no, and that reasoning matters more than the rule itself.

It's no surprise then, that no major publisher, ethics body or university accepts an AI system as an author. That's been true since it first hardened in early 2023. Authorship is a formal credit with legal and ethical responsibility for a published work; it's more than who typed the sentences. The reasons against AI being an author don't have anything to do with whether it's a good writer or how much of the paper it wrote. They come down to two specific obligations built into what authorship means. A language model can't meet either one.

TextPulse's own map of university and journal AI policy covers this question as one line inside a much wider picture of what institutions and publishers allow. This piece stays narrower and slower: the specific criteria that make authorship impossible for a model, quoted from the two bodies that wrote them.

Can AI Be an Author? What ICMJE's Four Criteria Require

ICMJE, the International Committee of Medical Journal Editors, defines authorship through four criteria, and a contributor has to meet all four to qualify, not just one or two. The last two are where a language model runs into a wall.

1. Substantial contributions to the conception or design of the work, or to the acquisition, analysis, or interpretation of data for the work
2. Drafting the work or reviewing it critically for important intellectual content
3. Final approval of the version to be published
4. Agreement to be accountable for all aspects of the work, ensuring that questions about its accuracy or integrity are investigated and resolved

A model can plausibly touch the first two. It can generate text that resembles a substantial contribution, and something that reads like drafting. The third and fourth criteria are where the category breaks down. Final approval requires a judgment about whether the finished paper is ready to stand behind. Accountability requires being able to answer for it afterward, under questioning, over years, if a result is challenged or an error is found. A language model has no continuity across that timeline and no capacity to be held to it.

What ICMJE Says About AI and Authorship

ICMJE's own recommendations state the conclusion in one sentence: Chatbots (such as ChatGPT) should not be listed as authors because they cannot be responsible for the accuracy, integrity, and originality of the work. That single clause does the same work as the four criteria above, compressed: responsibility is the thing a chatbot cannot hold, so authorship is the status it cannot occupy. ICMJE does not stop at forbidding the byline. Where a study used an AI tool, the same recommendations ask for disclosure: use should be reported in the acknowledgment section, and the human authors remain answerable for anything the tool contributed.

What COPE Says About AI and Authorship

COPE, the Committee on Publication Ethics, reaches the same place by a different professional route. Its position statement holds that AI tools cannot meet the requirements for authorship, because they cannot take responsibility for the submitted work. As non-legal entities, they cannot assert the presence or absence of conflicts of interest, and cannot manage copyright and license agreements. Authorship, in COPE's reasoning, is more than an intellectual credit. It is a legal and administrative role, and a chatbot cannot sign anything, hold anything, or be sued over anything.

ICMJE and COPE, Side by Side

ICMJE and COPE land in the same place, but they arrive by different professional routes, one editorial and one ethical. Seeing the two side by side is the fastest way to check a specific claim against its source.

	ICMJE	COPE
What it is	International Committee of Medical Journal Editors, sets authorship criteria for medical journals	Committee on Publication Ethics, sets ethics guidance for publishers across every discipline
Position on AI authorship	Chatbots should not be listed as authors because they cannot be responsible for the work	AI tools cannot meet the requirements for authorship because they cannot take responsibility for the submitted work
Why, specifically	Cannot be responsible for the accuracy, integrity or originality of the work	Cannot hold copyright, manage a license, or declare a conflict of interest
What is required instead	Disclosure in the acknowledgment section; the human authors stay responsible	Disclosure in the body of the paper, in both the Abstract and the Methods section

The Papers That Tried It Anyway

The clearest test of a rule is what happened before it was enforced. In January 2023, at least four medical journal articles briefly listed ChatGPT as a co-author, credited by name in the byline rather than mentioned as a tool in the methods. One of them, an editorial on AI platforms in nursing education published in Nurse Education in Practice, named ChatGPT alongside its human author when it first appeared.

Shortly after, the journal published a corrigendum saying that ChatGPT could not be listed as a co-author due to the journal's own guide for authors and its publisher's publishing ethics policies (the same reasoning ICMJE and COPE apply). The credit moved out of the byline and into an acknowledgment, which is where the guidance above says any AI contribution belongs. Every other 2023 case followed a similar path once an editor or a reader noticed: correct the record, keep the paper, move the credit down the page.

What to Do Instead of Listing AI as an Author

A tool that helped with a paper does not have to disappear from it. The credit just moves to a different place: name the tool and version, describe what it was used for, and confirm a human reviewed and takes responsibility for the result, in the acknowledgments or methods section rather than the author list.

TextPulse's own policy on ethical AI use in research and coursework draws the same line: using a tool is one act, and crediting it as though it did the thinking is another.

Where a discipline or a journal expects the AI tool cited rather than just named in a sentence, a dedicated AI citation generator formats that entry the way a reference list expects, without implying the tool did anything more than what the acknowledgment already says it did.

Publisher rules on this diverge in the detail, and journal AI policies are compared side by side separately. University policies are collected in their own piece for anyone submitting coursework rather than a manuscript.

The next paper that tries to credit a model as an author will meet the same four criteria, and fail the same two. The rule is not going to soften as the tools get better, because better output was never the thing authorship required in the first place.