This year, the amount of AI data has more than tripled, with all industries jumping on the bandwagon like crazy. But to be honest, no matter how advanced the model is, it will still fail when faced with the hurdle of "data fraud." If the training data fed to AI is unreliable to begin with, even the most powerful algorithms are useless—this industry-level bug has actually already attracted attention, and some people are working on solutions.