It's not something that can be enforced. For instance, the Unity Asset Store sells many AI-generated assets (sound effects, music, textures...), and it's often hard to tell whether an asset was AI-generated because many assets aren't disclosed as such.

Another gray area would be AI-generated world posters that users can submit (patreon posters etc.)

How's anyone going to tell whether the Udon code was AI generated, or whether someone used Photoshop's gen-AI tools.

Sometimes, it's difficult to tell whether something was AI-generated or not, and false reports can happen

I think this is important as to state as neuturally as possible there are users who dont want to support AI, and there are some who do. To give each type of community member the ability to sort that for themselves would be a huge benefit.

I would even be fine with OPs suggestions of being able to filter it out to not be there as incase VRChat also feels like it could create bias. But as long as I know what has been used before going into world and report worlds that are clearly inaccurate I would be onboard.

I'm leaving another comment here to boost this post and to add some more talking points for the devs. I think that this declaration should be required if it's used in any part of the world info page or in any of the assets. The same should be true for avatars. It needs to work the same as the comfort and safety tags. Generated a world description? Declare it. Generated the thumbnail? Declare it. Generated code for the world? Declare it. Generated any textures or models? Declare it. The rules for this should also clearly state a difference between generative AI and NPCs/procedural terrain generation. Obviously you wouldn't use this declaration on a world that procedurally generates terrain chunks or a dungeon level, but some people get confused with these things, so it's best to have it stated in the rules.

I would much prefer if generative AI was banned outright from the platform, but given the lack of action this whole time I feel like they've made their choice. I personally will not be giving VRChat another penny until they take action.

There's too many worlds with AI generated assets and thumbnails that eat up space in the world list and just about all of them have nobody in them and usually don't make it past community labs because people don't want them. The ones that do make it past almost always have sexually suggestive thumbnails and aren't marked as suggestive, which makes me worried that kids are the ones going to those worlds.

You should also be able to report worlds that fail to disclose.

True markets are informed markets. That's why nutrition labels are a thing.

I don't believe VRChat can in good conscious say this sort of labelling doesn't matter: AI is the hottest button in content creation, they have a label for way less important things, and Steam does this.

The only reasons I can think of for not doing this put profits over people: the pure revenue generation from increased monetized output, investor contract requirements, or plans to endorse/create AI content or systems themselves.

If anyone can think of any reason that doesn't put profits over an informed market, please let me know. Especially VRChat. Non-action would speak volumes for itself, though.

AI Police, show me your files.

Steam does it already and VRChat should do it too.

Setting the option to filter it, and that creators will need to declare it, will set a standard going forward for AI content that the online world badly needs. At the moment AI is too invasive and uses too much of others content without permission.

At the very least disclosure should be a thing. Being able to filter it would be nice but disclosing that something was made with generative tools should probably be implemented as these tools unfortunately gain more and more traction.

I know of at least three major world creators who pad their maps with generative slop and skyboxes and don't disclose it at all. The fact it's so normalized is frankly disappointing since currently none of these tools have proper regulations in place to protect the creators they're scraping from.

Is this even a problem? How many AI content in VRC?