What is AI art doing to illustration and stock photo work?

In a few months, AI image generators have gone from research demos to tools anyone can use. DALL·E 2 dropped its waitlist in September, Midjourney runs an open beta on Discord, and Stable Diffusion's model files can be downloaded for free. Type a sentence and you get a set of finished-looking images almost immediately. Since July, OpenAI has given DALL·E users full rights to sell and reprint the images they make.

For people who earn a living from pictures, that raises an urgent question. Illustrators, concept artists and stock photographers are seeing clients experiment with these tools, and the companies that sell images are trying to decide what the tools mean for their business and their legal exposure.

What the image marketplaces are doing

The clearest signal came in September, when Getty Images, one of the largest stock image agencies, banned the upload and sale of illustrations made with tools such as DALL·E, Midjourney and Stable Diffusion. Getty's CEO, Craig Peters, told The Verge the reason was legal uncertainty: questions about who holds the copyright to AI outputs, and unresolved rights in the images the models learned from, including the people shown in them. The models were trained on pictures gathered from across the web, and some generators even reproduce a smudged version of the Getty watermark.

Smaller art communities, including Newgrounds and FurAffinity, had already banned AI images, FurAffinity saying such content undermines human artists. The Verge reported that Shutterstock, Getty's big rival, seemed to be restricting certain searches for AI images without having adopted a formal ban. So for now, the major platforms are moving cautiously rather than embracing AI images.

Peters also argued that this is not the first technology to flood the world with pictures. Digital cameras and smartphones did that too. In his view, a stock agency's value lies in helping customers find the right image, not in how hard images are to make.

What illustrators are worried about

For individual artists, the concern is more personal. MIT Technology Review reported that the Polish fantasy illustrator Greg Rutkowski, known for work on games such as Dungeons & Dragons and Magic: The Gathering, had become one of the most used names in Stable Diffusion prompts. One site that tracks prompts counted his name about 93,000 times, far more than Picasso or Michelangelo. When Rutkowski searched for his own name online, he found images labeled with it that he had never made. He told the magazine that AI art is "starting to look like a threat to our careers."

Artists say they risk losing paid work when clients can generate something "in the style of" a known illustrator for free. San Francisco illustrator Karla Ortiz described a coalition forming across artistic professions to push for new policies. Right now, artists have no way to opt in or out of the datasets these models learn from, though a new site called Have I Been Trained lets them search to see whether their work is included.

A worked view: tasks at risk and new openings

No one has solid numbers yet on lost income, so the table below is a way to think through your own work, not a forecast. It sorts common image jobs by how easily a generator can stand in for them today.

How can illustrators adapt?

Start by understanding the tools rather than ignoring them. Spend an afternoon with DALL·E or Stable Diffusion and see for yourself where they excel and where they fall short. Test the things your clients need: readable lettering, the same character across ten images, an exact technical diagram, a revision that changes one detail and nothing else. Knowing the results firsthand lets you explain to clients, in concrete terms, what they would gain by hiring you.

Lean into what a generator cannot provide. Clients pay for a clear concept, revisions that follow feedback, a style that stays consistent across a whole project, and images they can use without legal doubt. That last point matters more now: with Getty refusing AI images over rights concerns, original work with clean ownership has a selling point it didn't need to advertise before. Keep your sketches and working files, because they show how a piece was made.

Some artists are also using the tools as assistants, for thumbnails, color studies or reference, while keeping the final work their own. Whether that fits depends on your field and your clients, and it is worth being open with them about how you work. Finally, watch the policy front. Groups like the Content Authenticity Initiative, which includes Adobe and Nikon, are building standards to record where an image came from, and artists' coalitions are pushing for consent and credit. The rules are still being written, and artists who stay informed will have a voice in how they turn out.