I don't follow the logic. One such article getting published does not automatically translate to no editoral oversight on any generated content.

Even a low percent of all published articles containing such problems doesn't in any way prove there's no editor involved.