Who owns an AI render? The authorship question nobody has answered

A visualisation produced by prompting a diffusion model sits in an unresolved legal space. The practice believes it owns the image. The client assumes the same. Neither has usually read what the tool's terms actually say.

Three jurisdictions, three answers

The US Copyright Office has held that purely AI-generated images lack the human authorship required for protection, while allowing that substantial human arrangement may qualify. The UK retains a provision for computer-generated works with no human author. The EU position remains unsettled.

Why this matters at contract stage

Most architectural appointments assign copyright in drawings and visualisations to the practice, then license to the client. If the image is not protectable, that clause transfers nothing.

A practical position

Document the human contribution: the massing model, the prompt iterations, the compositing. It is the strongest available argument that the image is arranged rather than merely generated.