As the old saying goes, never ask a woman her age, a man his salary or an A.I. company what they used to train their datasets. It’s not often discussed anymore, but the entirety of A.I. is built on copyrighted material, and the overwhelming majority of that was used without the copyright owners’ permission.

This, to most people, seems like a violation. After all, if a musician samples another artist’s song, they have to pay them a portion of the proceeds. However, for some reason, Large Language Models have looked at this system and claimed that they can just take your stuff and, well, that’s it.

Now, OpenAI has responded to concerns about copyright by saying that it’s going to keep using your copyrighted materials — unless you ask them to stop.

According to the Wall Street Journal, OpenAI has been telling talent agencies and studios that, if they don’t want their stars and films included in Sora’s training data, they have to tell them. Otherwise, they’re just going to use it — including, it seems, copyrighted characters.

To make things more frustrating, the WSJ says that OpenAI isn’t even allowing people to do a blanket opt-out. Instead, copyright holders will be given a link to report any violations of their copyright that they find.

How about this — just don’t use it in the first place!