Who Owns an AI-Generated Image?

With a few words of instruction, image-generating AI can now produce pictures in the style of a watercolor, an oil painting, or a movie poster in seconds. The results can be striking, and they have sparked a heated debate about creativity, ownership, and fairness in the art world.

One question is whether such images can be protected by copyright at all. In the United States, the Copyright Office has taken the position that copyright requires human authorship. A work created entirely by a machine is therefore not eligible, although a person who selects, arranges, or significantly modifies AI output may be able to claim protection for their own contribution.

A second question concerns the data used to train these systems. Many models learned from millions of images collected from the internet, including works by living artists who were never asked for permission and received no payment. Several artists and media companies have filed lawsuits, arguing that this amounts to large-scale copying. AI developers counter that learning from publicly available images is similar to how human artists study the works of others.

Courts in different countries may reach different conclusions, and new laws are being discussed. Meanwhile, some platforms have introduced tools that let creators opt out of training datasets, and others label AI-generated content so that viewers know what they are looking at.