Printmaking Keeps the Process in the Mark
Priya N. Chen on 3rd Nov 2026
Printmaking Keeps the Process in the Mark
I trust a print because the resistance shows. A line shifts between passes. Ink settles unevenly. The surface keeps every decision the artist made against the material.
That is the argument running through the Boston Printmakers 2026 North American Print Biennial, on view through November 21 at Boston University’s 808 Gallery, Tuesday to Saturday, 11am to 5pm. More than 600 artists submitted over 2,000 prints. Esther Adler, curator at the Museum of Modern Art, chose 120. The room holds a near-photographic black-and-white cityscape, a screenprint with acrylic paint and a trailing ribbon, and green guardian lions made with a LEGO tray. Etching, lithography, woodcut, linocut, screenprint, digital printmaking, and artists’ books share the walls.
The human point of view comes from Deborah Cornell, chair of printmaking at BU’s College of Fine Arts and cochair of the show. “One thing AI doesn’t have is hands,” she says. “These prints are made from drawings. They’re made from observation, made from life, made from being.” Joshua Brennan, the other cochair, shows Subcutaneous Field. He printed a colorful image, built a physical curtain, and printed a second version of the same image onto it. Foreground and background become hard to tell apart. Two passes, two surfaces, one hand lining them up.
The AI point of view has to be told as mechanism. An image model takes a prompt, applies patterns fixed during training, and samples a picture. Nothing in it presses, wipes, or misregisters a sheet. It does not want anything and it cannot sign anything. It is a system that a company built, trained, and owns.
So who is the author? In a print, the answer stands in the room: the artist whose hand left the marks. With a generated image, the candidates are the company that owns the model and the person who typed the prompt. The machine cannot claim rights. Only its owners can, and they do it through the terms they set. Those terms, not the pixels, decide whether the user owns the output.
The dispute starts earlier, at training. A model does not draw from nothing. Training adjusts millions of numerical weights until the system reproduces the statistical regularities of a dataset, and much of that dataset is prior art: prints, drawings, paintings, and photographs made by people who were not asked and were not paid. The weights are the company’s asset. The images that shaped them belong to someone else. Artists sued Stability AI, Midjourney, and DeviantArt in Andersen v. Stability AI on exactly this ground. The case is still in discovery in a California federal court.
The companies answer that learning from images is transformative and therefore fair use. The U.S. Copyright Office, in its May 2025 report on training, did not pick a side. It found that training on a large, varied dataset can be transformative, that copying whole works usually weighs against fair use, and that harm to the market for the original work counts heavily. It left the question to the courts and pointed toward voluntary licensing. The report is not binding, and the Office released it as a pre-publication version.
Output has its own ruling. In Thaler v. Perlmutter the D.C. Circuit held in March 2025 that copyright requires a human author, and the Supreme Court declined to hear the case in March 2026. A system cannot hold the copyright. How much human direction turns a prompt into authorship is still open. So the user’s claim is thin, the company’s claim runs through contract, and the artists whose work taught the model hold a grievance but, so far, no payment. In the 808 Gallery, every one of those links is visible in the mark.
Then there is the money. Selection is the first market event: 120 prints out of more than 2,000 is the filter that lets an edition carry a price, and the artist sells a physical object, edition by edition. A generated image can be copied without limit at no cost to whoever holds it. Its value rests on who holds the rights and on the fees and conditions the company attaches to the model. I would rather own a mark that cost the artist something than an image that cost the owner of the model nothing.
Join Neoclassical Pop Art daily take on the arts, the weekly letter, and first word on product launches, sales, and free gifts.