As a studio that used to use AI BG for a few of their games, but recently made a decision to stop..... yeah, this is not good.

โพสต์ดั้งเดิมโดย thatguythere03:

I was expecting a unilateral ban on anything with AI generated content/assets. What gives?

cheque cleared

The problem is that you're relying on people to self-report accurately. Adobe has already run into this honor system not working with the VERY recent Wacom tablet scenario. I strongly suggest walking this back until 1. LLMs can be properly vetted, as right now at least one contained CSAM and probably more will as well and 2. You develop a system that holds developers accountable outside of their own reporting because AI "creators" (that's not the right word for them but that's what I'm calling them in this case) have been proven on thousands of occasions not to be honest about the origins of their work.
Edit: I mean a system that doesn't rely on users to report AI usage, because let's be honest, that is NOT and should not ever be their job. Steam, YOU make the decision of what to let onto this platform, do not hold players responsible when this inevitably backfires. Same could be said of all content let onto this platform, really; please perform more rigorous checking or hire more people to accommodate this issue.

We need AI anti-fraud on steam

♥♥♥♥♥♥♥ disgusting. hope you people reconsider.

Disappointed that the outcome isn't more strict, sticking the guidance that self-assessment of content and signed statement from Devs that "you promise Valve that your game will not include illegal or infringing content" is very weak. This strikes me as self protection, whereas Steam should really be asking "how do you guarantee AI derived content is based on safe, legal, and ethically sourced training data". A lot of the big AI models can't guarantee that, and that's where the real risk is to both Valve and the developers.

Selling game with AI generated content is like asking yourself to be sued. Don't you all know about the viral article where the Open AI CEO basically admit that they used stolen and copyrighted material to train their AI?
Tagging system won't work; there's just too many cases already where people just don't tag their stuff as AI-made. And how do you (Valve) make sure the system where player can report ai-generated content won't be used to abuse developer?

disgusting.

I do not want AI content on my store feed, or anywhere on here whatsoever.

Extremely disappointed in Valve. AI is built on art theft, and both have absolutely no place on a platform where real games/art created by real people are being sold and distributed.

โพสต์ดั้งเดิมโดย Pakaku:

Extremely disappointed in Valve. AI and art-theft have no place on a platform where real games/art created by real people are being sold and distributed.

art theft is copyright infringement and is therefore already against the steam distribution agreement.

Like, if that's the rule you guys decide that it's good enough, then why not just ban them?
Nowadays every single AI-generated images, texts, voices, are based on massive copyright infringement, hell, even CEO of OpenAI just straight up admitted it.

Also, tagging never works, especially when you are dealing with liars who'll use AI-gen'd stuff anyway (for example one of the reason why wacom used AI-gen'd image in their ads, and get burned)
And don't even let me start with the nonsense like "We "trained" our model with our asset" which is always LoRA based on SD or pre-trained model.

Gross.

More trash and stolen assets incoming I guess?

Seems to fit the Epic Store more than Steam nowadays.

โพสต์ดั้งเดิมโดย Lixy:

More trash and stolen assets incoming I guess?

Seems to fit the Epic Store more than Steam nowadays.

Stolen assets are copyright infringement, which is against the steam distribution agreement.