Feature Requests

Just today I have seen 3 X/Twitter posts of fully AI generated worlds made using GPT-6 Astra, one of them already published. https://x.com/yuipxq/status/2097025460295389601 https://x.com/kurotori4423/status/2097011323767034098 https://x.com/pichikyo/status/2096285284178788612 I fear this kind of content will soon flood the platform. VRChat needed to add some kind of AI tag or descriptor a long time ago. This kind of quickly and effortlessly generated content hurts existing creators on this platform. Please don't let this become another "too little, too late" situation. VRChat needs an identifier for AI generated content added ASAP. Regardless of how end users personally feel about the use of AI, a way to separate and identify AI generated content from user created content is an absolute necessity.

Related to: https://feedback.vrchat.com/feature-requests/p/tackle-the-misleading-world-thumbnails-issue I think the main problem with AI-generated world thumbnails is not AI itself, but the fact that a thumbnail can be completely disconnected from what the world actually looks like. Currently, the only reliable way to find out is to download the world and enter it. For larger worlds, that can mean spending significant time and bandwidth just to discover that the thumbnail was misleading. I propose adding automatically generated in-world screenshots to the World Details page, separate from the creator-provided thumbnail. When a world is submitted to Community Labs, or when a new public version is published, VRChat could load the world and automatically generate preview images from inside it. The important distinction is that these images would be generated by VRChat rather than uploaded by the world author, and clearly labeled as such. 〇 Capture locations At least one screenshot should always be automatically captured from the world's spawn point. World creators could then optionally specify several additional capture points in the SDK to showcase important parts of their world. ・Automatic screenshot from the spawn point: required ・Additional creator-defined capture points: optional This gives creators some control over presenting the interesting parts of their world, while still giving users an objective view of what they will actually see when they first join. These previews could be regular screenshots, or potentially 360-degree captures. 〇 Advantages This would not require users or moderators to decide whether a thumbnail is misleading. Unlike a rating or review system, it would also avoid issues such as subjective moderation or malicious downvoting. Creators could still freely use artwork, renders, logos, or AI-generated images for their main thumbnail. Users would simply have an additional way to compare the creator-provided promotional thumbnail with an automatically captured view of the actual world before downloading it. 〇 Possible manipulation Of course, this would not make deception completely impossible. A creator could build an area specifically for the automatic capture, or attempt to detect the capture process and manipulate the scene using shaders or Udon. However, the goal does not need to be perfect anti-cheat. Requiring at least one capture from the spawn point, combined with additional viewpoints or 360-degree captures, would make manipulation significantly more difficult than simply uploading an unrelated thumbnail. Even if the system can theoretically be bypassed, a clearly labeled "Automatically captured in-world" preview would still give users much more information about a world before they download it.

Content that uses generative AI should have to declare that it is being used, and should be able to be filtered out of search results and recommendations by the user.

Issue With the raise of AI there's a growing issue that's hard to go unnoticed - many worlds that use AI image generation (but not just them) for their thumbnails, often mislead their users about their actual content (with the focus on quality) of the world. As a world explorer the problem is impossible to ignore at this point and increasingly annoying. I'm not talking about having to perfectly represent or show the exact photo of the content of the world in the thumbnail - a stylized or somewhat responsibly reimagined representation is ok - but the thumbnails should not mislead you into believing that this is what you will get that would leave you disappointed. I'm talking thumbnails that make you believe that it's an actual in-world photo e.g. showing you a breathtaking high detailed beach but upon enter it's a sand and water planes with low poly palms. Platform moderation The obvious issue is defining where the line is and the pure subjectivity of it. With moderation it'd be easy to go overboard and severely limit the freedom of world thumbnails and it would also greatly increase the work for the T&S team. There's a certain creativity with the relationship between the world page and the actual world. The obvious cases where someone is consciously and maliciously misleading you about the world's content or outright trolling you should be moderated tho. Community judgement Best idea would be to offload this judgement to the users e.g. in the form of already suggested world voting. A world that gives you high hopes but leaves you disappointed would be represented by the community vote. Worlds tab & algorithms Right now I feel like these "trap" worlds especially gain traction because once they pass a certain threshold they will end up in the "New & Noteworthy" tab and will pretty much get stuck there and have their "baitness" amplified even more. The Worlds tab algorithms are just something to consider too in the process of tackling this issue somehow. Better representation Another idea would be to allow multiple photos of the world and e.g. have the game automatically take an extra photo of the world or allow the author to replace with one that MUST be originating from within the world. Final though This is not exactly a specific feature request, nor duplication of the already existing "Add world voting" tickets, but casting limelight to the - I believe - increasingly visible problem that affects the quality of the platform and the joy of world exploration.

With the wave of generative AI content used throughout games/platforms that run on user-generated content such as VRChat, it should be required that creators tag their avatars & worlds appropriately with "Contains Generative AI Content" when the avatar/world contains ANY generative AI assets/content. The goal is so that users who do not wish to see such content, can filter it out through VRChat's native Content Gating system. Screenshot below as an example.

The title is pretty self explanatory, a filter for groups, avatars, profiles, and worlds if they have an AI generated image. How this feature would work is the player can choose to hide (or inform about) content that contains AI-generated media, if the content scanned for AI-generated media, a bot icon would appear in the corner of the thumbnail or instance. This ensures if a player would not like to see AI-generated media, they can use this feature.

Changing the language in the game interface, changing the keyboard language. - Adding a button on the game keyboard to change the language, it will be convenient for vr owners to type in the chat.

Allow the person that drops the portal to enter it first, delaying everyone else, so the person that dropped the portal becomes the first player to join the new instance. This would be handy for game worlds where the first player becomes a master of the instance, though also I believe this could be handy in some club instances as well. Example: if I drop a portal, allow me to enter through it first - other people will be able to join after a short delay (like, 10 or 15 seconds) to allow me to get to the instance first.

For as long as it's valid (not outdated or unmaintained), keep the option to claim the stickers and see the information for the crisis line. Don't have it tied to the marketplace as an expiring listing, since people won't be able to claim it after it expires, but rather have it be "it's own thing", if you know what I mean. For example, maybe a permanently available redeemable thing since it's free. Or, redeemable from some special information page related to the crisis line and mental health resources. My reasoning is that having it presented as a temporary store item kind of contradicts the supposed premise, which is to spread awareness of the crisis line affiliation.

Kicks last 1 hour and can't be undone. Please add an option to remove an active kick from the group moderation tools. It's needed when someone gets kicked by mistake, or when a kick is appealed and the moderators agree to lift it. Bans can already be undone, so kicks should be too.