Packshots and photos without a model no longer carry a wording about a model. If your label says something about the model in the photograph, such as AI MODEL, it now appears only on photos that have one. Packshots, and photos from sessions where no model was chosen from the casting, get a second wording instead.
You will find it under Wording on photos without a model, below Wording, on the AI label on exported photos screen in your team settings. It is shown only while your wording describes a model, and it offers only wordings that describe no one. Until you choose, it uses a matching wording that keeps what your label says about the product: AI MODEL becomes AI, for example, and a wording stating that the product is real keeps stating it.
The preview on that screen shows both: the white-background packshot with the second wording, the scenes with your main one.
The next labelled download of a packshot or a photo without a model is prepared afresh, because its label changes. If your wording does not describe a model, nothing changes for you: the same label on every download, and the same files.