> And people might not understand that their stories will influence the story of others, so... They don't. Pretty much all large-scale AI models separate infer

> And people might not understand that their stories will influence the story of others, so... They don't. Pretty much all large-scale AI models separate inference/content generation from training. What you use is the frozen version which doesn't update unless its creators retrain it. And if they retrain then it's their choice which data to use for retraining, there's no need to restrict the generated output.