No - when an image is generated using a diffusion model, the model is recognising patterns of pixels assigned to keywords across a very large dataset. This allows it to produce extremely realistic images but any resemblance to a real human being is purely coincidental.