It will. It's no different than Wikipedia quoting news articles that were sourced (uncredited) from Wikipedia. Circular referencing isn't new to AI.
Yeahhh... This is a huuuuuge problem with psychology research already. There are so many studies that get cited in textbooks and influence the training of actual therapists that, if you trace them back, go through a bunch of different references and mayyyybe eventually a small survey or a study of like 34 grad students. But a lot of what gets published is people interpreting things from sources that referenced another source that was created because of some common source.