The intentional or unintentional manipulation of maps has always existed, but the US government has recently given it greater attention and awareness. On the other hand, the basic principles of cartography - modeling, coding and communication - per se represent an inherent risk of falsification of maps. This aspect is illustrated in this contribution using examples from cartographic generalization - data classification and changing size. This also leads to the question of suitable training and test data for AI-generated map creation. Possible solutions such as the most automated, detailed metadata generation possible are presented.