It also how you define "fake", eg in machine learning it can be everything from fabricated results to data leakage, low quality data, bugs, p-hacking, weak base

It also how you define "fake", eg in machine learning it can be everything from fabricated results to data leakage, low quality data, bugs, p-hacking, weak baseline design, insufficient handling of randomness, etc etc.