Using AI in Research Papers: Journal Rules You Must Follow

Yes, you can use AI tools while writing a research paper, but most journals now attach clear conditions: an AI tool cannot be listed as an author, you must disclose how you used it, you remain fully responsible for every word and citation, and reviewers must not upload manuscripts into AI tools. This guide explains these rules and gives you a disclosure template you can adapt.

Why journals regulate AI use

Generative AI tools can draft text, summarise sources and suggest structure. They also produce confident errors, invent references that do not exist, and reproduce patterns from their training data. Journals are not banning the tools; they are making sure that a human author stays accountable and that readers know how the work was produced.

Rule 1: AI cannot be an author

Authorship carries responsibility: an author must approve the final version, answer for the accuracy and integrity of the work, declare conflicts of interest and sign copyright or licence agreements. A software tool can do none of these things.

The Committee on Publication Ethics (COPE), whose guidance is widely followed by journals, has taken the position that AI tools such as ChatGPT or other large language models cannot be listed as authors of a paper. COPE’s reasoning is that these tools cannot take responsibility for the work and, as non-legal entities, cannot declare conflicts of interest or manage copyright and licence agreements.

The same logic applies to acknowledgements that read like authorship. Do not describe an AI tool as a “co-writer” or credit it with the paper’s ideas.

Rule 2: Disclose how you used AI

COPE’s position also says that authors who use AI tools in writing a manuscript, producing images or graphics, or collecting and analysing data must be transparent about it, stating which tool was used and how, in the methods or a similar section. Journals implement this in different places: a methods paragraph, an acknowledgement, or a separate declaration at submission.

What usually needs disclosure

• Drafting or rewriting passages of text.
• Generating summaries of literature that you relied on.
• Creating figures, images or tables.
• Coding, data cleaning or analysis.
• Translation of substantial text.

What often does not need disclosure

Many journals do not require disclosure of basic spelling and grammar checks of the kind built into word processors. Policies differ, so read the author guidelines. When in doubt, disclose. Nobody has been penalised for being too transparent.

Disclosure template

Adapt this to your actual use:

Declaration on the use of AI tools: During the preparation of this manuscript, the author used [name of tool] to [specific purpose, for example improve the language of sections 2 and 3 / generate a first summary of the case law, which the author then checked against the original judgments]. The author reviewed and edited all output and takes full responsibility for the content of this paper.

Be specific. “AI was used” is not a useful disclosure; “used to suggest a structure for Part III, which was then rewritten” is.

Rule 3: You are responsible for every word

COPE’s position stresses that authors are fully responsible for the content of their manuscript, including any part produced by an AI tool, and are liable for any breach of publication ethics. In practice this means:

• Every fact must be checked against a reliable source.
• Every legal proposition must be checked against the statute or judgment. AI tools often confuse repealed and current provisions, for example mixing up sections of the Indian Penal Code, 1860 and the Bharatiya Nyaya Sanhita, 2023.
• Plagiarism rules still apply. If AI output closely reproduces someone’s published text or idea, you are the one who plagiarised.

Rule 4: Never submit fabricated citations

This is the single biggest risk for law students. AI tools can produce case names, reporter citations and journal articles that look perfect and do not exist. Courts in several countries have criticised lawyers for filing such citations, and for a journal it is a matter of research misconduct.

A verification routine

1. Treat every AI-suggested source as unverified until you have opened it yourself.
2. For cases, find the judgment on the court website or a recognised database and confirm the name, year, reporter citation and the paragraph you rely on.
3. For statutes, check the current text on India Code.
4. For articles and books, find the actual publication, not just a listing.
5. Delete any source you cannot locate. Do not “fix” it by guessing.

For research workflows that use AI safely, see our guide to AI tools for legal research in India.

Rule 5: Confidentiality in peer review

If you review papers, remember that an unpublished manuscript is confidential. Uploading it to a public AI tool to summarise or critique it may share the author’s unpublished work with a third party. Many journals prohibit reviewers from doing this, or require permission and disclosure. Write your review yourself; if you use a tool to polish the language of your own comments, check whether the journal allows it. To understand reviewer duties more broadly, read what is double-blind peer review.

Images, data and AI detection

• Images: many journals restrict or prohibit AI-generated images in research papers unless the images are the subject of the research. Check before you use one.
• Data: never use an AI tool to create or “fill in” data. That is fabrication.
• AI detection tools: some journals and universities run them. Their results are not proof, but heavy undisclosed use creates avoidable suspicion. Writing from your own notes is the best protection.

A sensible way to use AI in a law paper

The lower risk uses keep your thinking at the centre. The higher risk uses either hand the thinking to the tool or invite fabrication.

Checklist before submission

• No AI tool listed as author or credited with ideas.
• Disclosure statement written, specific and placed where the journal asks.
• Every citation opened and checked by you.
• No AI-generated data or undisclosed images.
• Author guidelines read for the journal’s own AI policy.

Also see how to avoid plagiarism and co-authorship rules, because the same accountability principles apply to human co-authors.

Frequently asked questions

Can I use ChatGPT to write my research paper?

You can use it as an aid if the journal allows, but you must disclose the use, verify everything it produces and take full responsibility for the content. Submitting AI-written text as your own unaided work breaches most journal policies.

Can an AI tool be listed as a co-author?

No. COPE’s position is that AI tools cannot meet authorship requirements because they cannot take responsibility for the work, and journals following COPE apply this rule.

Where do I put the AI disclosure?

Wherever the journal’s author guidelines say: commonly the methods section, the acknowledgements or a separate declaration at submission.

Do I need to disclose grammar checking software?

Many journals do not require disclosure of basic spelling and grammar tools, but policies vary. If the tool rewrote substantial text, disclose it.