Hacker News new | past | comments | ask | show | jobs | submit login

I don't think it's really about enforcement. It's more about liability.

Valve has an official position that they don't allow AI content (apparently). When the lawsuits show up they can say that they don't serve any AI content as official policy. When someone points out the AI content that they do serve, then they pull out their expert witness that testifies that their AI detection method is as good as possible and they couldn't haven been expected to do any better. Meanwhile, they're more than happy to remove anything explicitly flagged that falls through the cracks.

Finally, I suspect that anyone who can prove that they're able and willing to indemnify Valve against lawsuits for AI content that their game contains will be allowed to have AI content.




> Finally, I suspect that anyone who can prove that they're able and willing to indemnify Valve against lawsuits for AI content that their game contains will be allowed to have AI content.

Yes, they're quoted as saying that AI generated assets are permitted if the developer can "affirmatively confirm" they own all the IP in the training set. seems reasonable to me.


Yeah, I saw that quoted part, although, I suspect that if I show up with a bunch of AI assets that I can prove are 100% mine, then the reviewer is likely to error on the side of Valve not being sued.

Meanwhile, AAA blockbuster studio will almost definitely be given a pass with a wink and a handshake after saying, "Hey if anyone figures it out, we'll take the blame." For using assets that throw up multiple red flags.




Consider applying for YC's Spring batch! Applications are open till Feb 11.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: