That wouldn’t work for this - people save them to their camera rolls and post directly. It’d need some kind of image recognition / OCR and heuristic, which feels wildly like too much effort for what is basically a mild annoyance.