Disclosing AI Use in Art Listings and Exhibitions
An AI disclosure is a short, factual statement about how a work was made. It sits in a marketplace listing, on a gallery wall label, or in the paperwork an artist submits before a show. The wording looks simple, but the decision behind it is not. Artists worry that admitting machine involvement will cost them sales, while buyers and selection panels want to know what they are actually looking at. What follows covers where disclosure belongs, what it should say, and how to write one that is honest without being self-defeating.
Why disclosure sits at the centre of AI art practice
Disclosure puts the audience in a position to make a choice. Mark Koester’s AI Included Disclosure Acknowledgment, published in June 2024, makes the case directly: AI usage disclosures allow people to actively decide to support and consume art made exclusively or primarily by AI, or by humans. That framing treats the statement as a service to the viewer rather than a confession from the maker.
There is a protective argument too. Commentary on the subject argues that the use of AI, particularly for imagery, is a real and present danger to the careers of many people in the creative industry, and that using AI is a choice while disclosing its use should not be optional. From that position, silence shifts risk onto buyers, illustrators and galleries who have no way of telling how a piece was produced.
One commentary frames the question around disclosure of true and accurate facts about a work of art, including who created it and whether the creator was non-human or artificial. That is a useful test for any statement you draft. Read it back and ask whether a stranger would come away with an accurate picture of the person, and the system, behind the image.
What marketplaces and exhibitions actually ask for
Requirements vary by venue, but the pattern across published policies is consistent: disclosure is expected at the point where the work is presented, not buried somewhere else. Saatchi Art’s AI-generated art policy states that if AI tools were used at any stage of the creation process, artists must disclose this in the description of their work. Exhibition practice follows a similar logic, with artists submitting AI-assisted artwork required to complete an AI Process Disclosure as part of their submission and exhibition label documentation.
| Setting | What is typically required | Where it appears |
|---|---|---|
| Online marketplace listing | Disclosure in the description if AI tools were used at any stage | Work description |
| Exhibition entry | Completed AI Process Disclosure | Submission paperwork and exhibition label |
Two details in that wording deserve attention. The first is the phrase at any stage. A disclosure is not only for images that came out of a generator fully formed. If a model touched the initial sketch, a texture pass, or the final enlargement, that belongs in the statement. The second is the label. Exhibition labels are read by people standing in front of the work, often before they read a catalogue entry, so the wording needs to work in a single sentence.
Policies change, and different platforms set different bars. Always check the current policy of the marketplace, gallery or competition you are dealing with rather than relying on a summary written months earlier.
What a good AI disclosure contains
A workable disclosure answers the questions a buyer or juror would ask if they were standing next to you. You do not need to publish your full workflow, but the statement should cover the essentials.
- The tools or systems used, named plainly rather than referred to vaguely as digital enhancement.
- The stages where they were used: concept, composition, image generation, upscaling, colour work or clean-up.
- The extent of the contribution, for example whether a model produced a base image later repainted by hand, or assisted only a small element.
- What you did yourself afterwards, including hand painting, collage, printing, framing or digital correction.
- Any prompt or process notes the venue specifically requests as part of its submission.
Notice how much of that is about sequence. The most common failure in a disclosure is not dishonesty but vagueness, where the reader cannot tell whether AI generated the whole image or helped with one layer. A process sentence fixes that faster than any amount of justifying language.
Writing the statement for a listing description
Listing descriptions are read quickly and often on a phone. Lead with the process, keep the sentence short, and place the disclosure where it will not be missed rather than in a footnote. A structure that works: name the tool used for the base, describe the human work on top, and state what the buyer receives.
For example, you might write that a base composition was generated with a named tool, then repainted, restructured and colour corrected by hand in a named program, with the final image combining generated and hand-worked elements. If the piece is a physical print, say what was printed and from which file. Avoid soft phrases such as digitally created where a model was genuinely involved, because they read as concealment once discovered.
On your own portfolio site you have more room. A short line under each work, linked to a fuller process note, keeps the gallery experience clean while giving the curious reader everything they need. Where a marketplace provides a description field, use it, but confirm the platform’s current policy first so your wording matches what it asks for.
The reputational question, and why artists hesitate
Hesitation is rational. Research reported in February 2026 found that disclosing AI use harmed an employee’s reputation, and that explicitly stating AI had not been used did not improve evaluations either. That second finding is the more interesting one, because it suggests audiences are wary of the topic itself rather than simply punishing honesty. The study concerned employees rather than exhibiting artists, so treat it as a signal rather than a verdict on your practice.
What the finding argues against is the bare statement. A line that says only that AI was used invites the reader to imagine the worst-case version of your process. A sentence describing which parts were generated and which were hand-worked gives them something accurate to judge, and it demonstrates that you know your own workflow in detail.
Research published in June 2026 analysed 23 AI-mediated artworks, discussing 13 of them closely to show distinct approaches to technical disclosure. The practical point is that there is more than one route: a plain written statement, a documented technical record of prompts and edits, or both. Venues increasingly want the second, while buyers respond to the first.
Thresholds, percentages and the question nobody can answer
Community discussions keep circling a threshold. Is 50 percent AI the point at which a work can no longer be called hand drawn or painted, or is it 20 percent? Artists also ask about a 30 percent rule. Nothing in the sources behind this article establishes a fixed percentage, and no agreed figure emerges from the debate.
Rather than hunting for a number, describe the process stage by stage. If a venue does ask for a proportion, answer as precisely as you can and explain what the figure refers to, because a percentage of the image area is a different measure from a percentage of the working time. Honesty about the measure matters more than the number itself.
A practical disclosure checklist
- Read the current policy for the marketplace, gallery or competition before you submit or publish.
- Record your process as you work, including tools, versions and the stages where a model was involved.
- Draft one sentence for the label and two or three for the listing description.
- State the human contribution clearly, including any hand work done after generation.
- Complete any AI Process Disclosure form in full rather than leaving fields blank.
- Review older listings periodically, since a statement written for one policy may no longer match another.
Disclosure works best as a habit rather than a one-off task. Artists who document as they go find that the label writes itself, and that the conversation with a buyer becomes a discussion about craft instead of a defence of the file.
Frequently Asked Questions
Do I need to disclose AI use in my art listings?
Often, yes. Saatchi Art’s policy, for example, requires artists to disclose in the work description if AI tools were used at any stage of the creation process, and exhibition submissions commonly require an AI Process Disclosure. Rules differ between platforms and venues, so check the current policy for each one before you publish or submit.
What should an AI disclosure actually say?
Name the tools used, state which stages they were involved in, and describe what you did by hand afterwards. The aim is an accurate picture rather than a defence. A reader should be able to tell whether a model generated the base image or helped with one element, and what human work sits on top of it.
Is there an agreed percentage threshold for calling a work hand drawn?
No fixed figure emerges from the published discussion. Artists ask whether 50 percent or 20 percent is the line, and a 30 percent rule is sometimes mentioned, but no agreed standard appears in the sources. Describing your process stage by stage is more useful than quoting a single number, and clearer for buyers.
Does disclosing AI use damage an artist’s reputation?
Research reported in February 2026 found that disclosing AI use harmed an employee’s reputation, and stating that AI had not been used did not improve evaluations. That study looked at employees rather than artists, so treat it as a caution. Context helps: a specific process statement reads very differently from a bare admission.
What is an AI Process Disclosure?
It is the documentation some exhibitions require from artists submitting AI-assisted artwork, completed as part of the submission and used in the exhibition label. In practice it records how the work was made and where machine systems were involved. Ask the venue for its exact form and complete every field rather than leaving gaps that invite guesswork.
