Should Writers Disclose That They Used AI?

Whether to disclose AI use depends on how much of the finished piece it produced, not whether it was used at all. If AI only helped with research, disclosure isn’t necessary. If it drafted passages a person substantially reworked, most readers don’t need a label. If AI produced most of the finished text with only light editing, tell your readers.

Separate The Question From “Did You Use AI At All”

Almost every writer working today uses AI for something, research, a fact-check, an idea they wanted to argue with. Disclosing that level of use to readers would be like disclosing which search engine you used, technically accurate and practically meaningless. The real question isn’t whether AI touched the piece anywhere. It’s whether AI, rather than a person, is responsible for the words the reader is actually reading.

The 3 Tiers That Actually Matter

AI helped with research, structure or a first idea, and a person wrote the finished piece: no disclosure needed, because a person is the author in every sense a reader would care about. AI drafted meaningful passages and a person substantially reworked them, checking every claim and rewriting for voice and argument: disclosure is a courtesy rather than an obligation, and most publications don’t provide it, because the finished piece reflects a person’s judgement throughout. AI produced most of the finished text and a person’s involvement was mainly reviewing it: tell your readers. That’s the line where a reasonable reader would feel misled if they later found out, and a misled reader is a reader you don’t get back.

Internal Disclosure Is A Separate Decision

Whatever you decide about readers, your own team should always know how a piece was made. An editor signing off on a piece needs to know whether they’re checking a person’s first draft or an AI’s, because the checking required is different in each case. Conflating internal transparency with reader-facing disclosure is a common mistake, and it usually ends with teams either over-disclosing to readers out of caution or under-disclosing internally out of habit.

What A Reader-Facing Disclosure Line Actually Looks Like

Where disclosure is warranted, it doesn’t need to be defensive or lengthy. A single line at the top or bottom of a piece, stating plainly that AI produced an early draft which was then edited and fact-checked by a named person, does the job. Readers generally aren’t hostile to AI involvement itself, they’re hostile to discovering it was hidden from them. A short, matter-of-fact disclosure line tends to cost a publication far less trust than an evasive one does once a reader works it out on their own.

When In Doubt, Ask The Honest Question

Would a reasonable reader feel misled if they found out exactly how this piece was made, in the amount of detail you’d be comfortable giving them? If the honest answer is yes, disclose. If the honest answer is that they’d shrug, because a person clearly stands behind the piece regardless of what helped them get there, you don’t need to.

What This Looks Like In Practice

A trade publication might decide that opinion columns always carry a person’s name and are never disclosed, because the writer is visibly accountable for every word regardless of what helped them draft it. The same publication might decide that short, templated market updates, where AI drafts most of the finished text from data feeds with only light human review, always carry a line stating as much. Neither decision is more “honest” than the other in the abstract, they’re honest because the team thought about what a reader would actually want to know and wrote the answer down, rather than defaulting to whatever felt least awkward in the moment. Writing the rule down in advance also removes a harder conversation later, the one where a single writer has to decide, alone and under deadline, whether this particular piece is the exception. That decision, made under time pressure by one person with a deadline bearing down on them, is exactly the moment a sound policy tends to quietly bend.

This is one of five questions answered in the full guide, AI Should Draft, Humans Should Craft: Where The Line Sits In An Editorial Team.

If you’re watching all of this happen and wondering how to make AI work for your business, book a call with me today – I train and consult businesses and teams to use AI with confidence, skill and intention.