@gamingonlinux I love what Valve is doing here, but I don't like the title of the article nor the caption of this post. Valve is not banning AI tools, it requires proof of the datasets copyright that was used to teach the model. This is the right way I think.

@gamingonlinux fucking based, common valve W

@gamingonlinux I feel this is has to be a form of automated detection and may lead to false positives, much like YouTube.

@gamingonlinux The thing that amazes me most about this is that humans at Valve seem to be intimately involved in this specific case. The article finishes on an ominous note: a hands-on approach from Valve's seems unlikely to scale. That said, Valve has already been operating at scale for a long time. So to me this case is a positive sign that they ARE still able to address issues like this with humans, since it is such a rarity these days (YouTube of course being the opposite)

@gamingonlinux but why? What if that’s the whole point of the game, creating a game using AI?

@gamingonlinux Honestly, as someone who likes to play around with stuff like Stable Diffusion, this seems like a perfectly reasonable stance from Valve. The flagrant disregard of copyright has always been a huge issue with these AI models, and taking the stance of "prove ownership of the assets used to make the model" is a good CYA policy until the law finally catches up.

@gamingonlinux well its nice to know that i dont have to spend time making a game, no point in making a game if i have to make it without textures.
I cannot draw and learning to be good takes so much time that its too late to start now.

@gamingonlinux i know it's just valve covering their own asses legally, but still incredibly based