Summary

A German higher court ruled that turning a copyrighted underwater dog photograph into a comic‑style image using AI does not constitute copyright infringement. The judges said the photographer’s protected expression – composition, perspective, lighting, and depth of field – was not copied, only the underlying concept of a dog lunging for a toy, which is not protected. The ruling came after a separate dispute involving stock photographer Robert Kneschke and the LAION dataset, where the court found the dataset’s use fell under non‑commercial scientific research.

The decision hinges on traditional copyright doctrine that shields the specific expression of a work while allowing the reuse of ideas. AI tools can replicate the feel of an image by altering enough visual details, satisfying the court’s “idea‑expression” test. This narrow, case‑by‑case approach gives AI developers more leeway to train on and remix existing photographs without fear of outright infringement.

For photographers, the ruling threatens revenue streams tied to licensing and exclusive rights, while AI firms gain a clearer path to build models on existing imagery. The legal landscape is still evolving, with German courts signaling they will decide each AI‑copyright case individually rather than issuing broad statutes, leaving the industry in a state of flux.