Text-to-Image AI (Image Generation)
Text-to-image AI creates an image from a written description – one that is not photographed but computed by models trained on large quantities of image-text pairs. That does not mean the result resembles no protected original: research has shown that diffusion models can reproduce individual training images almost identically.
In practice
Text-to-image AI works well for concept images, mockups and some marketing visuals, but it should be labelled transparently (see AI labelling obligation) and checked for quality and image rights. For genuine team photos or other shots where authenticity carries trust and E-E-A-T, it does not replace real photography.