AI spits out exact copies of training images. A paper that was published where they found that it's been coming to question what extent do these image models memorize things versus include novels. It turned out that you can exit samples with with prompts and yeah this is a kind of big finding in a way and points to maybe there are more concerns with commercializing these things if that's Yeah, and I guess we're seeing a lot of that with a jail breaking of chat GPT as well right where people have been to say what its prompts were and and that sort of thing so like this idea of data extraction.