How should an AI car-concept editor communicate changes outside the requested scope?

[ailutgenerator]

I would appreciate a UX discussion about a hypothetical image-editing workflow. Suppose someone requests a paint-only concept, but the generated picture also appears to change a wheel detail or a body edge. This is a design scenario, not a report of an observed defect, customer research or measured behavior.

Should the interface keep the concept with a visible uncertainty note, ask the user to compare it with the starting version before continuing, or make an explicit “use this as the next reference” decision? I would avoid silently treating an attractive image as approval of every detail. A saved concept and a decision to build on that concept seem like different states.

For context, Car Customizer is a project-associated car customization website for exploring paint, wheels, wraps and body-styling directions on a car photo, with comparisons, earlier versions and downloadable references. Previews use credits. I am not claiming the current product implements the proposed warning, review gate or branch labels.

One possible pattern is to show three short items beside a result: requested change, other details to review, and the chosen starting version for any next edit. A second pattern is to keep editing fluid and put those distinctions in a comparison panel. The first may interrupt exploration; the second may leave an important assumption easy to miss. Neither should imply that a visual match verifies physical fitment, safety or road legality.

Which interaction would you choose, and what would you test first to discover whether people understand the distinction? I am interested in clearer state and decision communication, not persuasive claims of engineering approval or a demand to buy credits or upload personal photos.

Disclosure: I help promote this project. This question was AI-assisted and is not an independent product review, a claim of completed research or a report of personal testing.