Citing Large Language Models in Academic Writing: Standards and Best Practices
Large Language Models (LLMs) can help brainstorm, outline, and polish text; but in academic work, they raise questions about authorship and citation. Many students and researchers still struggle with when to treat an AI tool as a source versus when it simply supports the writing process.
This guide explains when to cite, when to disclose, and how to format references for tools like ChatGPT using style-guide updates from the APA, MLA, and Chicago formats. It also covers how universities and journals expect AI use to be acknowledged to maintain transparency and academic integrity.
<CTA title="Write With Proper Attribution" description="Use Jenni AI to structure sections and keep your citations consistent across styles." buttonLabel="Try Jenni Free" link="https://app.jenni.ai/register" />
What counts as “use” that requires citation or disclosure?
Understanding when to cite or disclose an AI tool depends on how deeply it shaped your work. Let’s break down the main scenarios so you can stay compliant without over-citing.
When do you cite vs only disclose?
If you quote or paraphrase specific LLM output, you cite it. But if you only used it to brainstorm, outline, or refine grammar, most venues expect a disclosure instead, usually in your acknowledgments or methods section.
Example:
“We used ChatGPT to organize preliminary ideas, but all writing and analysis were performed by the authors.”
Can an LLM be an author?
No; AI tools cannot take authorship credit. The Committee on Publication Ethics (COPE) and the ICMJE state that authorship requires accountability and consent, qualities only humans can provide.
Think of AI like a calculator: it supports your reasoning, but it doesn’t own the results.
Why disclosure matters even if you do not quote the output
Even if the LLM’s words never appear in your paper, disclosure helps maintain transparency. Elsevier and other major publishers now require authors to clarify what tools were used, how they were applied, and how accuracy was verified.
In short: Transparency shows responsibility; silence creates doubt.
<ProTip title="🧠 Pro Tip:" description="If you are unsure whether to cite or disclose, disclose. Journals prefer over-transparency to unclear authorship." />
Core principles before you format anything
Before you start formatting citations, understand what keeps AI references credible and reproducible.
Attribution and verifiability
Readers and reviewers must clearly see which parts of your text used AI help. Proper attribution builds transparency and accountability.
🧩 Example:
“This paper used ChatGPT to refine grammar and structure, not to generate core arguments.”
For a stronger grasp of why verification matters, check Nature’s report on AI verification which discusses traceability and responsible model use.
Model versions evolve. Should you capture version and date?
Yes. Because LLMs evolve, include model name, version, access date, and prompt context so your reference can be verified later.
✅ Key inclusions:
Model name
Version or release
Access date
Prompt context
<ProTip title="📘 Reminder:" description="Record the model name, version, and access date when generating output so your citation stays reproducible and audit ready." />
Formatting LLM Citations Across Major Styles
Every major citation style has adapted differently to AI tools. Understanding these changes helps you write citations that stay both transparent and credible.
APA 7th Edition — the most structured model
APA treats an AI model like an authored software tool. Include:
Developer as author (e.g., OpenAI)
Year of release
Model name and version in brackets
URL and access date if available
🧩 Example (APA format):
OpenAI. (2024). ChatGPT (March 2024 version) [Large language model]. https://chat.openai.com
MLA 9th Edition — focusing on version accuracy
The MLA Style Center encourages listing both the model name and version in the Version element, with the prompt context if you quoted a generated response.
🧾 Example (MLA format):
OpenAI. ChatGPT, version GPT-4, OpenAI, 2024, https://chat.openai.com.
Unlike APA, MLA prioritizes reproducibility, ensuring readers know exactly which version produced the text.
Chicago Style — flexible but specific
Chicago allows AI citations in both notes and bibliographies. Each entry should name the developer, model version, generation date, and URL if stable. The Chicago Manual of Style Online explains that URLs for transient models should be omitted, while Georgetown University Libraries provide note-based examples.
<ProTip title="📚 Writing Tip:" description="In Chicago, prioritize permanence. Include a URL only if it will stay live or can be archived reliably." />
IEEE — waiting for formal rules
IEEE hasn’t yet published a universal format for AI citations. Most institutions temporarily mirror APA or Chicago until an official template is released.
AMA and medical writing — emphasizing transparency over format
The AMA Manual of Style urges authors to explain how AI was used rather than list it as a citation. This aligns with medical publishing’s preference for disclosure statements over formatted references.
Transparency guidance from OUP Academic and ICMJE confirms that human authors remain responsible for all AI-assisted content, from drafting to interpretation.
How Major Publishers Handle AI Use
Leading publishers are rapidly updating policies around AI use in research and writing. While all require transparency, their approaches vary in tone and enforcement.
Science — zero tolerance for AI-generated text
The Science family prohibits AI-generated text or images without prior approval. Their editorial standards on Science.org highlight that all text must be authored by humans, and any use of LLMs must be disclosed.
Example disclosure:
AI tools were used to refine the language of this manuscript. The scientific content was written and verified by the authors.
<ProTip title="🚫 Submission Reminder:" description="Science requires human authorship for all text and figures. Always disclose AI use to avoid rejection during editorial screening." />
Nature — emphasizing accountability
The Nature group focuses on accountability and transparency rather than punishment. According to their updated ethics statement. AI tools can assist writing, but cannot hold authorship or intellectual responsibility.
Elsevier — formalized and consistent
Elsevier unified its AI guidelines across journals through its central framework. Authors must declare AI use clearly during submission and describe its purpose in improving language, analysis, or structure.
✅ Common disclosure elements:
The tool’s name and version
How it assisted (editing, outlining, data formatting)
A statement of human oversight
<ProTip title="📄 Author Tip:" description="Keep your disclosure short and factual. Mention how the AI assisted your work and affirm that all analysis and conclusions were human-led." />
arXiv and preprints — open but transparent
arXiv prohibits listing AI tools as co-authors but encourages clear statements of AI involvement in submissions. Because preprints shape early-stage research, arXiv prioritizes disclosure over restriction.
🧩 Example:
The authors used ChatGPT for initial phrasing suggestions. All ideas, revisions, and conclusions remain the authors’ own.
<ProTip title="🧠 Editorial Tip:" description="Preprint servers care more about transparency than format. Add a short acknowledgment of AI use instead of a formal citation." />
Citation Examples for ChatGPT and Other LLMs
These quick reference samples show how to cite AI tools in major academic styles. You can copy and adapt them to fit your paper’s formatting.
APA format
OpenAI. (2025). ChatGPT (GPT-4.1) [Large language model]. https://chat.openai.com
In APA style, include the developer, release year, model name and version, and a retrieval link if available. This format also applies when citing responses generated by other large language models.
MLA format
OpenAI. ChatGPT, version GPT-4.1, OpenAI, 2025, https://chat.openai.com.
MLA recommends listing the version number and access details for reproducibility. If the AI’s response is quoted directly, note the prompt in the text or Works Cited entry.
Chicago notes format
ChatGPT, GPT-4.1, OpenAI, response to author’s prompt on 7 November 2025, https://chat.openai.com.
In Chicago style, add the prompt date when relevant and include a stable URL only if the AI output is accessible or archived.
IEEE template
[1] ChatGPT, GPT-4.1, OpenAI, response to prompt, Accessed Nov. 7, 2025. [Online]. Available: https://chat.openai.com
IEEE has not yet released formal guidance, but most institutions follow a layout similar to this. Including access dates keeps your reference verifiable.
<ProTip title="📘 Quick Tip:" description="Always include the model name, version, and date of access so your AI references stay verifiable and consistent across citation styles." />
Common Questions About Citing ChatGPT and Other LLMs
These quick answers cover the most-searched issues around citing AI tools in academic work.
Should I cite the LLM or the sources it surfaces?
Cite the original sources whenever possible. If you’re quoting or paraphrasing text directly generated by an LLM, cite the model itself according to your chosen style guide.
Do I include my exact prompt in the reference?
Include your prompt only if required by the journal or format. APA allows prompt citations, while MLA and Chicago may suggest adding it as a note or contextual detail instead.
What if the model version changed between drafting and submission?
Add the model version and access date in your citation. If the tool was updated mid-process, note that version drift in your methods or acknowledgments section.
Can reviewers use LLMs to review my paper?
Most publishers discourage this. Journals like Science and Nature explicitly prohibit uploading manuscripts into public AI tools during peer review.
Is AI detection reliable enough to police this?
Detection tools are still inconsistent. Publishers emphasize responsibility and disclosure over reliance on AI detection systems.
<ProTip title="💡 Quick Tip:" description="When unsure, disclose. Clear acknowledgment of AI use builds trust with editors and readers more than hidden assistance ever could." />
Acknowledging AI Use with Jenni AI’s Declaration Feature
Jenni AI now includes an AI Declaration feature that helps authors add transparent statements about how AI assisted in their writing. By typing the command /AI Declaration in the Jenni editor, users can instantly insert a short, journal-appropriate acknowledgment that meets disclosure requirements from publishers such as Elsevier, Nature, and arXiv.
<CTA title="Add Transparent AI Acknowledgements" description="Use the new AI Declaration feature from Jenni to easily add a compliant AI usage statement to your paper." buttonLabel="Try Jenni Free" link="https://app.jenni.ai/register" />
Example output:
During the preparation of this work, the authors used Jenni AI to assist with language refinement and reference structuring. After using this tool, the authors reviewed and edited the content as needed and take full responsibility for the final version.
This declaration format has already appeared in several Q1 and Q2 journals, highlighting a broader shift toward responsible AI use in research.
You can see more examples on Google Scholar.
Writing Ethically in the Age of AI
Citing AI tools correctly isn’t just about format, it’s about credibility. Clear attributions show how technology supports your reasoning rather than replaces it, helping readers trust your process and conclusions.
<CTA title="Cite Smarter with Jenni AI" description="Use Jenni AI to structure citations, format references, and build academic papers that meet global publishing standards." buttonLabel="Try Jenni Free" link="https://app.jenni.ai/register" />
As citation rules evolve, the goal remains the same: transparency. When you show how you’ve used AI, you contribute to a more ethical, traceable, and open future for academic writing.