Cite a ChatGPT conversation in APA 7
ChatGPT and other AI conversation citations are new enough that APA 7 (2020) is still settling on a format. The current consensus is to treat the AI as the author, the version (e.g., GPT-4) as the year, and the prompt as the title, with a stable share URL if one exists.
APA 7 rules for a ChatGPT conversation
- Treat the AI as the author: OpenAI for ChatGPT, Anthropic for Claude, Google for Gemini.
- Use the model version (e.g., GPT-4, Claude 3.5 Sonnet) as part of the title.
- The prompt you used appears in the title or in square brackets as a description.
- Use the date of the conversation, not the model's release date.
- Include a share URL if one is available; otherwise note that the output is unreproducible.
- APA treats AI chats as personal communications; reference list entry is optional but recommended.
Worked example
APA 7 · ChatGPT conversationA real ChatGPT conversation formatted using the APA 7 rules above.
OpenAI (2024, November 2). Explain the difference between APA 7 and MLA 9 [Large language model]. *ChatGPT (GPT-4)*. https://chat.openai.com/share/example-share-id
Build your own APA 7 reference
Paste a URL, DOI, or ISBN below. The generator is preset to APA 7.
We’ll detect the type and pull metadata via CrossRef (DOIs), Open Library (ISBNs), or Open Graph tags (URLs). Edit anything before copying.
Citation fields
APA 7 ChatGPT conversation citation guide
How APA 7 decided to treat AI chatbots
Citing ChatGPT and other large language models is new enough that the guidance has shifted more than once, so it is worth knowing the current APA position rather than copying an old forum answer. APA treats the model itself as the author. For ChatGPT that author is OpenAI; for Claude it is Anthropic; for Gemini it is Google. The year is the year of the version you used, and the title is the name and version of the model, italicised, followed by a bracketed description of its type. The standard pattern is: OpenAI. (2024). ChatGPT (Mar 14 version) [Large language model]. https://chat.openai.com/.
APA's reasoning is that the model is the thing producing the text, so it occupies the author and title roles together, much as a piece of software does. This is why you do not put your own name or your prompt in the author slot. The version matters because model behaviour changes between releases, and a reference to GPT-4 in early 2024 describes a different system from the same name a year later.
The reproducibility problem and where the prompt goes
The deepest issue with citing an AI conversation is that it is usually not reproducible: type the same prompt twice and you get different output, and other readers cannot retrieve your exact exchange the way they can retrieve a journal article. APA's advice is to acknowledge this directly. Because the output is not recoverable, APA recommends describing how you used the tool in the text of your paper and, where it helps the reader, including the full prompt and the relevant portion of the response in an appendix rather than burying it in a citation.
If the platform offers a shareable link to the conversation, include that URL in the reference so readers can at least see your specific session. If there is no share link, the reference still ends at the tool's general URL. The prompt itself is best reported in your own running text — for example, noting that you asked the model to summarise a debate — because the prompt is part of your method, not part of the source's title.
In-text use and academic-integrity cautions
In text, an AI citation looks like (OpenAI, 2024). When you quote the model's output directly, present it as a quotation and attribute it, just as you would a human source. Crucially, check your institution's rules before you rely on a chatbot at all: many courses restrict or forbid AI assistance, and a tidy citation does not make unauthorised use acceptable. Citing the tool is about transparency, not permission.
Remember that language models confidently invent facts, fake references, and misremember details, so anything the model tells you must be verified against a real, citable source before it goes in your work. Use the generator below to format the OpenAI or Anthropic reference correctly, but treat the conversation as a starting point for finding authoritative sources, never as the authority itself. The safest pattern in academic writing is to cite the primary sources the model points you toward, with the AI reference reserved for cases where the model's own output is genuinely the object of study.