How to Cite AI Tools: A Verification and Disclosure Checklist

Citation and disclosure solve different problems

A citation helps a reader identify a source. A disclosure explains how a tool affected the work. Your institution may require one, both, or neither for a particular use. Grammar correction, idea generation, translation, generated text, and data analysis may be treated differently.

Start with the instructor’s policy and the current edition of the required style guide. Do not assume a citation generator knows the local rule.

Record the interaction while it is available

Save the tool name, provider, model or mode when shown, date, prompt, and a shareable URL or transcript if permitted. Note which part of the assignment it influenced. AI outputs can change, so a generic homepage link alone may not let a reader recover the interaction.

Cite original evidence, not the AI’s retelling

If a tool says a study found a result, find the paper and verify the population, method, result, and limitations. Cite that paper for the claim. A separate AI acknowledgment may describe how you discovered or summarized it.

Never preserve a fabricated reference just because it is formatted perfectly. Search the DOI, title, and authors independently.

Audit generated citations

Check author order, title, container, year, pages, DOI, capitalization, and citation style. Then confirm the source actually supports the nearby sentence. Citation generators reduce formatting work; they do not perform source evaluation.

Use a plain-language disclosure

A useful disclosure is specific: “I used [tool] on [date] to generate counterargument questions and receive clarity feedback. I verified factual claims against the cited sources and wrote the final text.” Adapt the wording to your institution’s policy and never claim a process you did not follow.