Im curious to see the progress on video. My mental model is that video is in many ways going to be harder to get right than text. If these models are fundamentally limited by their training data, then it will be really easy to get something impressive looking, but really hard to get it to do just what you want it to. Also, lots of risk of leaking specific IP.

Maybe a certain type of creator can design within these limits. And certainly lots of AI slop can come to Tik Tok. But if you have a specific artistic or brand vision mind it will be impossible to achieve.