As AI models are shared and reused, it matters to know how one was built and where it should not be trusted. A model card is a short, standardised document describing a model's intended uses, the data behind it, how it was tested and its known limitations.
Like a nutrition label, it helps someone decide whether a model fits their situation and warns of weak spots, such as groups it was not well tested on. Cards are only as honest and complete as their authors make them, and they do not fix underlying problems, but they encourage transparency and more informed, responsible reuse of models.