I, too, have received emails about articles or papers that I supposedly wrote, but which simply do not exist. It is a phenomenon that researchers increasingly encounter, as AI systems produce convincing-sounding but fabricated references. Through Inside Higher Ed, I discovered that these are now called Frankencitations. These citations appear to be compiled from existing authors, half-titles, real journals, and completely fictional details.

But honestly? I don’t think this is just an AI problem.

While investigating educational myths, we constantly encounter erroneous citations. Articles that misrepresent authors. Researchers attributed statements that cannot be found anywhere. References copied from paper to paper without anyone checking the original source. This is how we end up, for example, with the famous Maslow’s pyramid, even though that pyramid does not appear anywhere in his own work. Sometimes a study simply doesn’t exist. Sometimes it does exist, but it says something completely different from what everyone keeps repeating. This process is called citation laundering: questionable, erroneous, or unproven claims gain increasing credibility as people keep citing them without returning to the original source.

Often, that is even one of the biggest frustrations of our work. Not only searching for sources, but also finding out whether they really exist. And if they do exist, whether people have read them correctly.

AI may well accelerate and exacerbate that problem. After all, a language model can sound very credible and “academic.” But the breeding ground has been there for a long time. Citation laundering, source pollution, and academic phone games existed long before ChatGPT.

Therefore, in my opinion, frankencitations do not imply that technology suddenly introduces errors, but rather that they reveal how often science and education depend on trust in secondhand sources.

Or shorter: check the footnotes every now and then. Asking it for a friend.