We have been writing artist statements for years, and lately every draft runs into the same snag: how do we talk about the AI tools sitting somewhere in our process without sounding either defensive or vague. A line like "AI was used in this work" tells a viewer almost nothing. It does not say where, how much, or why, and it leaves the kind of gap that makes people assume the worst about the labor involved.

This piece is about building that credit the way we build anything else in the studio: by naming materials, naming stages, and being specific about what our hands did versus what a tool did. We work mostly in mixed media and digital print, so the examples below come from that practice, but the method translates to painting, sculpture, textile, or sound.

Why vague AI disclaimers don't help anyone

A one-line disclaimer such as "created with the assistance of AI" sits in a statement like an unlabeled jar of pigment. It might be cadmium, it might be a synthetic substitute, and the viewer has no way to know what they are looking at or how it behaves over time. Vagueness does not protect us from scrutiny, it just delays the questions and usually makes them sharper when they finally come.

We learned this the hard way at a small gallery show in 2022. A visitor asked whether the figures in one of our digital collages were generated or drawn. We said "a bit of both," and that answer satisfied no one, including us. The conversation that followed made clear that people were not objecting to AI use, they were objecting to not knowing what they were looking at.

Specific crediting does three concrete things a vague line cannot:

• It tells a curator or buyer which parts of the piece are reproducible versus unique, which matters for editioning and pricing.
• It gives collectors a sense of the work's materials in the same way "oil on linen" does, so they can judge longevity and handling.
• It protects us from the accusation of hiding something, because the credit reads as confident rather than apologetic.

None of this requires a confession-style paragraph. It requires the same discipline we already apply to listing gesso, graphite, or archival ink: name it, place it, move on.

Deciding what actually needs crediting

Not every brush touch needs a footnote, and not every AI touch does either. The test we use is simple: if removing the tool would change the final image, text, or object in a way a viewer could notice, it gets credited. If removing it would only change our workflow and not the outcome, it usually does not.

We sort our own use into three rough tiers, and we credit differently depending on the tier:

A practical example from our shelf: we scan charcoal drawings at 600 dpi, then run a denoising pass through an AI-based tool before printing. That is transformative. The drawing itself, every mark, came from a stick of vine charcoal and our wrist. We credit the denoising step in one short clause because it affects the printed surface quality, but we do not treat it as a co-author.

Naming the tool and the stage of use

Generic phrasing like "AI software" is the equivalent of writing "paint" instead of "gouache." It is technically true and practically useless. We name the specific tool, the version if it matters, and the stage of the process where it entered, the same way we would note that underdrawing was done in 2B graphite before the ink wash went on.

Here is the structure we default to, built from three parts:

1. Tool name, written plainly, no brand flourish ("Midjourney v6," "Adobe Firefly," "a locally run Stable Diffusion checkpoint").
2. Stage, meaning where in the sequence of making it was used (sketching, compositing, upscaling, color grading, text drafting).
3. Degree of reliance, meaning whether the output was used directly, edited, or only as a reference that we then redrew by hand.

So instead of "AI was used to help create the backgrounds," we write: "Background textures were generated in Midjourney v6 at the compositing stage, then painted over by hand in oil to merge them with the foreground figures." That sentence does more work in twenty-four words than most disclaimers do in a full paragraph, because it answers the three questions a viewer actually has: which tool, which part, and how much human hand followed it.

Sample process notes from our own statements

We keep a running file of credit lines we have used, partly so we are not reinventing phrasing for every show, and partly so we can see how our own disclosure has matured. Below are three real examples, lightly trimmed, from statements written between 2023 and this year.

From a 2023 print series: "Initial color palettes were explored using an AI image generator as a sketching tool. None of the generated images appear in the final prints; they served as reference only, the way a thumbnail sketch would. Final compositions were built from scratch in Procreate and printed on cotton rag paper."

From a 2024 textile piece: "Pattern repeats were generated through a custom-trained AI model built on our own archive of hand-block prints, then laser-cut into the resist stencils used for the indigo dye. The model saw only our prior work, no external dataset, and the output was edited by hand before cutting."

From this year's statement, covering a hybrid sculpture and video piece: "The spoken narration was drafted with help from a large language model for pacing and word choice, then rewritten in our own voice and recorded by us. The model did not generate the ideas or the research behind the piece, only assisted with sentence rhythm during editing."

Each of these notes names the tool, says where it sat in the sequence, and states plainly what it did not do. That last part, the negative statement, turns out to matter as much as the positive one. Viewers tend to fill silence with the most dramatic possible assumption, so closing that gap directly saves everyone time.

Where to place the credit in a written statement

We place AI credit in the same spot we would place any material note: inside the process description, not in a separate disclaimer paragraph bolted onto the end. A statement that reads like a terms-of-service page does not build trust, it just signals anxiety. If the tool use is integral, it belongs in the sentence that describes how the piece was built, right next to the paint, the paper, or the clay.

For a short wall label or caption, where space is tight, we use a compressed version: tool name and stage only, no degree-of-reliance clause. For example, a sixty-word caption might read: "Mixed media collage. Background elements generated in Adobe Firefly, composited and hand-painted." That is enough for a gallery wall. The longer version, with the full three-part structure, goes in the fuller statement available in a handout or on a website.

For grant applications and gallery submission forms that ask directly "was AI used in the making of this work," we answer at the tier level described earlier, generative, transformative, or administrative, and then refer the reader to the fuller process note in our statement. Some funding bodies have specific disclosure requirements that go beyond what we would volunteer in a wall label, and those forms should be read carefully; when the language of a grant or contract is unclear about what counts as AI assistance, it is worth asking the program officer directly rather than guessing, since the consequences of a wrong guess can affect eligibility.

Updating credits as your workflow changes

Our tool use from two years ago is not our tool use now, and the credit lines have to keep pace. We review our standard process-note language roughly every six months, usually right before a show deadline forces us to write a fresh statement anyway. That review has caught outdated lines twice: once where we had stopped using a generative step entirely and the old credit was simply inaccurate, and once where a new transformative step, an AI-assisted frame interpolation for video work, had quietly entered the workflow without ever making it into writing.

A simple habit helps here: keep a dated log, even a short one, of which tools touch which stage of a body of work. We use a plain text file with entries like "March 2024, added AI upscaling to print pipeline, 4x factor, applied after final color correction." When statement-writing time comes, that log becomes the source material instead of memory, which is unreliable under deadline pressure.

If a past exhibition's wall text or catalog entry becomes outdated because the tool changed or was removed, we do not usually go back and edit historical records, since that piece was made and shown under those conditions. We do make sure the current, active statement on our website reflects present practice, and we note the date of the statement itself so a viewer can tell how current the information is.

Common mistakes

The errors we see most often, in our own early drafts and in other artists' statements, cluster around three habits: naming the category instead of the tool ("AI" instead of "Midjourney v6"), describing the tool's presence without describing its boundary (saying what it did but not what it did not do), and burying the credit in a single line at the very end where it reads as an afterthought or a legal hedge rather than part of the material description.

A fourth mistake worth flagging separately: treating AI disclosure as a one-time task completed when the statement is first written, rather than as a maintained part of the studio record. Workflows shift faster than statements get rewritten, and a credit that was accurate eighteen months ago can quietly become misleading.

Next steps for your own statement

Start by listing every stage of your current process, from sketch to final finish, on paper or in a plain document. Mark which stages involve a tool with generative capability, which involve transformative processing, and which are purely administrative. Then write one sentence per generative or transformative stage using the tool-stage-degree structure outlined above, and fold those sentences into your existing material description rather than isolating them.

If your work involves commercial licensing, grant funding, or academic exhibition, it is worth having a knowledgeable colleague, gallery director, or, for contractual questions, a lawyer familiar with intellectual property review your disclosure language before submission, since requirements vary by institution and are still shifting as policies catch up with practice. Keep your process log current, revisit your statement every few months, and treat the credit line the way you would treat any other material note: specific, placed where it belongs, and honest about its edges.