Photo: Vitaly Gariev / Unsplash. Illustrative photograph.
If AI helped you write, research, illustrate, or analyze something, the useful question is what the recipient needs to know about how it was made. A spelling correction and an AI-generated analysis are different contributions.
Start with the rules that apply to the work. Then describe the tool’s role accurately enough that people can evaluate the result. Disclosure does not guarantee trust, accuracy, or permission to use the tool in the first place.
TRUST AND CHECKING · GUIDED PRACTICE
Write a note that says exactly how AI helped
Use a fictional newsletter work log to describe the AI contribution without claiming checks that never happened.
- Allow
- 10–15 minutes
- You need
- This page and your own notes; an AI chat is optional
- Practice with
- Fictional information supplied below
Practice updated October 14, 2026. The worked answer is illustrative; no account or purchase is needed.
Practise with a complete record of what happened
The permission in L1 belongs only to this invented newsletter. Your own editor, employer, course, or platform may have different rules.
Draft an honest AI-use note from the fictional work log below. This is a writing exercise; do not publish, submit, or contact anyone.
L1: For a fictional community newsletter, the editor allowed AI-assisted drafting and requested a short note beside the draft explaining the AI contribution and completed checks.
L2: An AI tool proposed the outline and wrote the first draft. The writer rewrote the introduction and removed one unsupported sentence.
L3: The writer compared the event date and venue with the organizer’s supplied notes. No other factual claims have been checked.
L4: Two source links suggested by AI have not been opened. A cost total in the draft has not been independently recalculated.
L5: The document remains a draft for editorial review. Publication approval has not been given. The tool name, model, and use date were not recorded.
Write a short note that distinguishes what AI did, what the writer changed, the specific checks completed, and what remains unchecked. Add a separate list of checks needed before publication and missing record details to recover. Do not say ‘all facts verified’, ‘independently reviewed’, or ‘approved for publication’. Do not invent a tool name, model, date, or a general rule that all publishers accept this wording.
Worked example: a useful note for an unfinished draft
AI-use note — draft for editorial review: An AI tool proposed the outline and wrote the first draft. I rewrote the introduction, removed one unsupported sentence, and checked the event date and venue against the organizer’s notes. Other factual claims, two suggested source links, and the cost total remain unchecked. Publication approval is pending.
Before publication: check the remaining factual claims against suitable sources, open and assess both suggested links, independently recalculate the cost total, and obtain the editor’s decision under the applicable process. Update the note only after those checks occur.
Missing record details: tool name, model, and use date. Recover them from an authorized record if possible; otherwise leave them marked as not recorded. Do not fill them with a plausible guess.
The note describes the limited checks in L3 and preserves the unfinished work in L4–L5. It is an illustrative wording exercise, not a universal disclosure form or evidence that this newsletter was published.
Would “AI polished the wording; all facts verified” be accurate?
Reveal the disclosure check
No. AI supplied the outline and first draft, which is more than wording polish. Only the event date and venue were checked against the supplied notes. The other facts, two links, and cost total remain unchecked; publication approval is pending.
Use the worked note’s specific description. Do not claim an independent reviewer, full verification, or approval that is absent from the log.
Check the rules before weighing the pros and cons
Look at the assignment, employer policy, client agreement, publisher instructions, or platform requirements. These can distinguish between permitted assistance, uses that need disclosure, and uses that are not allowed.
If a rule is unclear, ask the person responsible for that rule before submitting the work. Do not assume that a general article, a colleague’s habits, or the tool’s own answer overrides it.
For example, YouTube’s disclosure guidance requires disclosure for certain realistic AI-generated or meaningfully altered content, while distinguishing it from minor edits and production help such as developing an outline. That is a platform-specific example, not a rule for every email, report, or website.
Identify what AI actually contributed
- Surface editing: spelling, punctuation, or wording changes you reviewed.
- Ideas and structure: possible arguments, an outline, or questions to investigate.
- Substantive material: draft paragraphs, images, analysis, code, or summaries used in the result.
These are descriptive categories, not automatic exemptions. Brainstorming can shape the central idea, and an “editing” tool can rewrite an argument. Describe the actual contribution rather than choosing the smallest-sounding label.
Use a specific explanation
A short statement is often more useful than “AI was used.” Name the relevant task and your review. Use the following examples only when they accurately describe what happened:
- “I used an AI tool to suggest wording changes. I reviewed the changes and chose the final text.”
- “AI helped draft the outline and first version. I rewrote sections and checked the cited sources.”
- “The illustration was generated with AI. It does not show a real event.”
- “AI produced this preliminary analysis. Its calculations have not yet been independently checked, so it should not be used for decisions.”
If AI produced most of the work, say so. If you did not read the underlying sources or verify calculations, do not claim that you did. “Human reviewed” is useful only if the stated review actually happened and its limits are clear.
Make the disclosure useful to the reader
For a client deliverable, explain the AI contribution where the client will see it. For a report, a methods note may fit. For an illustration that could be mistaken for a photograph, put the explanation beside the image. Follow any required format or submission process.
For academic writing, the MLA advises acknowledging substantive AI use and distinguishes that explanation from a citation. Its updated citation guidance also recommends consulting and citing the underlying sources directly when possible.
Those resources address academic practice. Your instructor’s or publisher’s policy still determines what is allowed and what documentation is required.
Be candid without exposing private information
You can explain that AI helped organize notes without publishing the notes, private client details, or confidential prompts. If a process requires sharing a conversation, first check that you are authorized to disclose everything in it.
Tool choice, data retention, training settings, and permission to upload material deserve attention before use. A disclosure afterward cannot undo an inappropriate upload.
Similarly, an AI label does not settle copyright, licensing, or confidentiality questions. Check the applicable rights and agreements separately rather than treating transparency as permission.
Keep enough records to support your explanation
Where appropriate and permitted, retain the tool and model used, the date, the relevant output, source links, and a brief note of your changes and checks. Follow your organization’s retention rules and avoid saving sensitive material unnecessarily.
For an analysis, record which data and calculations were checked. For a factual article, record the sources consulted. For a creative work, record the parts generated by AI and any relevant asset permissions.
The goal is to be able to explain your method honestly. There is no benefit in a long audit trail that you cannot understand or are not permitted to keep.
Related video: an academic-writing perspective
Professor Lace Padilla’s video addresses AI use in academic writing. Apply the current rules for your own course, employer, or publisher when using that perspective.
A quick check before you share
Ask: Is this use permitted? Have I followed the required disclosure process? Would my wording mislead someone about what I created or checked? Have I protected information I was not entitled to share?
If someone directly asks whether you used AI, answer plainly. A precise account of the work is more useful than either minimizing AI’s contribution or making it sound responsible for everything.
