I suspect that many people will publish substandard generated data. This will affect the integrity of training data in the long term, unless steps are taken.