In short
Put a short label beside the AI-written text saying what wrote it, from what, and whether a person checked it, for example "AI-generated summary of the transcript, checked by Ana on 25 September". Mark sections a person rewrote, and keep the label when you copy the text.
To label AI-generated content in meeting notes, put a short line next to the AI-written text that says three things: that AI wrote it, what it was written from, and whether a person has checked it. For example: "AI-generated summary of the meeting transcript. Checked by Ana Ruiz, 25 September." Put it where the reader starts reading, mark any part a person rewrote, and keep the label when the text is copied somewhere else.
This guide gives practical wording for the places notes end up: an email, a document, a wiki page, a ticket, a chat message. What the law requires is a separate topic, covered in labelling AI meeting notes under the EU AI Act.
Label AI-written notes you share
- Decide what the label covers. Is the whole note AI-written, or one section? Label at the level the reader will read at: a section, a list, or the whole message.
- Write what produced it and from what. "AI-generated from the meeting transcript" tells the reader the source was the transcript, not someone's memory.
- Say whether a person checked it. Add a name and a date if you did. If nobody has, say "not yet checked".
- Mark edited sections. Where a person corrected or rewrote something, mark it: "edited by Tom".
- Put the label beside the text. At the top of the section, not in a footer the reader never reaches.
- Make sure it survives copying. Use plain text in the note itself, not only a tag, colour or icon that disappears when pasted.
What a good label says
A useful label answers the reader's real question: how much should I trust this?
| Part | Example | Why |
|---|---|---|
| Who or what wrote it | "AI-generated" | Tells the reader to read it as a draft |
| From what | "from the meeting transcript" | Tells them what to check it against |
| Checked by | "checked by Ana Ruiz" | Tells them someone took responsibility |
| When | "25 September 2026" | Tells them how current it is |
| Edits | "action items edited by Tom" | Tells them which parts a person vouched for |
Keep it short. One line is enough; nobody reads a disclaimer paragraph.
Wording for each place
In an email
Put it on the line before the notes, not in the signature.
In a document or meeting-notes page
A line under the heading of each AI section:
In a wiki or knowledge base
Pages outlive the meeting and get edited by others, so date the label:
In a ticket or task
Tickets are acted on, so the label should say what was checked:
In a chat message
Short, at the start:
Marking what a person checked
A label that says "AI-generated" on everything, forever, stops being read. What makes a label useful over time is showing which parts a person has since taken responsibility for.
- Checked, unchanged: "checked by Ana".
- Corrected: "edited by Tom" beside the part he changed.
- Rewritten by a person: you can drop the AI label for that part and say, if it helps, "written with AI assistance".
If you checked the notes against the transcript, say so; that is the step that makes them trustworthy. Can you trust an AI meeting summary? describes how to check.
Labels that do not survive
Some labels vanish on the way:
- Colour or icons only. Lost when pasted as plain text, and invisible to screen readers if not described.
- Footnotes. Separated from the text they describe as soon as someone copies a paragraph.
- A label on the page, not in the text. An app's badge next to a summary does not travel into your email unless the export carries it.
Plain words in the text itself are the only label guaranteed to arrive. When you export, check that the label came with it. Exporting meeting notes as Markdown covers export.
A simple rule for a team
If several people share AI notes, agree one rule so labels are consistent. For example:
- Everything an AI wrote carries "AI-generated" until a person checks it.
- Whoever checks it adds their name and the date, and changes nothing else about the label.
- Action items are checked before anyone is assigned work from them.
- Nothing AI-written goes outside the organisation unchecked.
- Labels stay on when text is copied into email, tickets and documents.
Five lines in a team wiki are enough. If your organisation has a policy on AI tools, the rule belongs in it.
Notes going outside your organisation
For a client, a partner or the public, the label matters more and the checking matters most. Strip anything said off the record, check every number and commitment, and send text rather than audio. Sending AI meeting notes to clients has a checklist for that case.
How Notey labels notes
Notey labels everything its AI writes, wherever it appears. This is a design decision, not a claim about any legal obligation.
- Each AI section is labelled "AI-generated": the summary, action items, minutes, tough questions, where you differed, the follow-up email, and Ask Notey's answers.
- Your edits are marked. Change an AI section and it is shown as edited by you.
- The labels go with exports. The summary, any note or everything exports as Markdown, to the clipboard or a file, with the AI and edited-by-you labels in the text. The transcript on its own exports as plain text.
Once the notes leave Notey, the label is in your hands. Keep it on when you paste, and add who checked it. How AI meeting notes are made explains why the check matters.
Frequently asked questions
How do I disclose that meeting notes were written by AI?
Add a one-line label beside the notes, such as "AI-generated from the meeting transcript; checked by [name]". Put it where the reader starts reading, and keep it when the text is copied into email or a document.
Do I have to label AI-generated meeting notes?
It depends on where you are, what you publish and your organisation's policy. For internal notes, labelling is usually good practice rather than a legal duty; the EU AI Act guide covers what Article 50 says. Check your own policy.
What if I rewrote most of the AI summary?
Say so. "Drafted by AI from the transcript, edited by Tom" tells the reader someone took responsibility. If you rewrote it completely, the label can say it was written by you with AI assistance.
Should the label say which AI model was used?
It can, but the reader usually needs something simpler - that a machine wrote it, from what source, and whether someone checked it. Your organisation may have its own rule.
Where should the label go in an email?
On the line before the notes, not in the signature or at the bottom. Readers skim the top, and a label at the end is often cut when the email is forwarded or quoted.