I used ChatGPT to help write parts of my research paper, and now my professor wants to make sure all sources are properly cited. I’m not sure about the best way to reference ChatGPT according to academic standards. Can anyone guide me on the correct citation style or format for ChatGPT in American English?
I’ve had the same question pop up when writing my last paper, and there was a bit of confusion since ChatGPT isn’t a traditional source like a book or an article. Most schools and journals are still figuring this stuff out so expect a bit of variation. For APA, their style guide (as of the 7th edition) suggests this: If you’re quoting or relying on content generated by ChatGPT, you cite it as “personal communication” because it’s not retrievable by others. In-text, it’ll look something like (OpenAI, personal communication, Month Day, Year). But, you DON’T put it on the reference list, just cite it in the text.
But, as of 2023/2024, some journals are starting to accept generative AI citations as proper references. For example, here’s how APA recommends you cite a ChatGPT conversation you can’t reproduce exactly:
OpenAI. (2023). ChatGPT (June 1 version) [Large language model]. https://chat.openai.com/
And then in text: (OpenAI, 2023).
If you’re using MLA or Chicago, their recommendations are similar. MLA has proposed:
ChatGPT. “Prompt you entered.” OpenAI, 13 June 2024, chat.openai.com/chat.
So, it’ll depend a lot on your professor’s preferences and what style manual you’re following. If your professor is super strict, I’d clarify with them, but the safe way is to treat it like personal communication for academic honesty, and include your prompts/responses as an appendix if you want to cover all your bases.
Bottom line: Mention it in your methods or acknowledgments if you used it significantly, cite as a personal communication unless your professor says otherwise, and always double check with whatever style guide you’re using since the “official” consensus is still shifting.
I’ll just say it: the “official” standards right now are a total moving target, and it’s almost funny watching journals scramble to catch up. @cacadordeestrelas nailed a lot of the current conventions (especially the thing about APA treating it like “personal communication”), but honestly, I think that approach is already getting dated, given how ubiquitous these tools have become.
Here’s my two cents. If ChatGPT actually contributed text or ideas to your paper (vs. you just using it to brainstorm or outline), the real academic honesty move is to:
- Explicitly disclose in your Methods/acknowledgments that you used AI assistance to draft or generate content—don’t just hide it in a citation.
- Include actual transcripts of your prompts and ChatGPT’s responses as an appendix or supplementary material. This way, anyone reading your paper knows exactly what you got from AI.
Citing it as “personal communication” feels weird when a billion people can freely access ChatGPT—it’s not an email from your grandma. Some disciplines and journals, especially in STEM, are already starting to want dated, versioned references like: “OpenAI. (2024). ChatGPT (GPT-4, June version) [LLM]. https://chat.openai.com/” in the bibliography, and then in-text citations as (OpenAI, 2024). This is cleaner, more transparent, and better for future readers to understand what exactly you used.
What bugs me is, a lot of profs still haven’t decided what they actually want. So, honestly, ask first, but err on the side of being overly transparent. And if you used AI to reword, edit, or fact-check, say so directly in the paper. Half the “standards” out there are just people making it up as they go—don’t stress, but definitely document. Not everyone agrees, though—I know some instructors who flat-out ban it or want you to treat it like plagiarism. Again: clarify, clarify, clarify.
And just for the record, I disagree with acting like ChatGPT is “personal communication” forever. In three years, that’ll sound as outdated as citing Wikipedia as “unpublished correspondence.” The times, they are a-changin’.
Let’s cut through the academic fog: “the right way to cite ChatGPT” is still a mess, but here’s the actionable stuff. Pros so far—citing ChatGPT (or any AI LLM) boosts transparency and avoids accidental plagiarism traps, especially now that instructors are hyper-vigilant. Cons—no universal standard, so you run the risk of irritating your professor or being at odds with (still-evolving) journal policies. Also, unlike regular sources, AI doesn’t “authentically” originate information, so it’s weird to cite it like a conventional author.
What the other folks pointed out—“personal communication” vs. full reference list entry—is accurate but not bulletproof. Being forced into the “personal communication” format just because ChatGPT’s responses aren’t perfectly reproducible makes sense today, but seems outdated now that AI tools are ubiquitous. That said, if you’re submitting to a journal with rules that are older than ChatGPT itself, you may be stuck with that workaround.
Real talk: If your work hinges on AI-generated text, drop an explicit note in your Methods or Acknowledgments (example: “AI-assistance using ChatGPT contributed to drafting/structuring certain sections.”) Append your prompts and outputs—full or partial versions—as supplementary material, so there’s no ambiguity over what’s “yours.” Harvard, APA, MLA, Chicago—they’re all scrambling, so whichever you use, document clearly and let your professor be the tie-breaker.
Compared to the other answers, I’d stress one extra layer: context matters. Did you use ChatGPT for basic brainstorming or for analyzing data? The more integral it was, the more prominent your disclosure should be. At the end of the day, you want readers to understand where the ideas come from, not just hit some citation checkbox.
Oh, and a quick rundown:
Pros – boosts transparency, prevents plagiarism accusations, aligns with the growing push for AI-usage disclosure.
Cons – inconsistency between guides, lack of permanence in AI outputs, possible confusion for readers down the line, professor/journal roulette.
Others covered most bases, but for what it’s worth, don’t hide your AI use—make it front and center if it shaped anything meaningful. That honesty will look way better than shoehorning ChatGPT into a dusty old format. The standards will change, but your academic honesty shouldn’t.