Article · AI translation
IT Home reported on October 11 that VRChat Inc., the developer of VRChat, announced its approach to AI-generated content on October 7 and updated its creator guidelines at the same time. The company explicitly prohibits the use of misleading AI-generated thumbnails for worlds and avatars, to prevent creators from attracting players' clicks with thumbnails that do not match the actual content at all.
VRChat said the platform has always regarded “freedom of expression,” “the boundless creativity of humans,” and “human connection” as core values. Although AI technology has improved developers' efficiency, it has also given rise to a large amount of low-quality content. In fact, low-quality worlds and avatars already existed on the platform before AI became widespread. The arrival of AI has further lowered the barrier to producing and publishing content in bulk, flooding the platform with substandard creations and making quality content more likely to be buried.
VRChat also revealed that its internal team is divided over AI. Some employees use AI to automate workflows or fill gaps in their knowledge, while others have deep reservations about it. Recently, an incident occurred on the VRChat website: AI-generated placeholder artwork originally intended for page layout was mistakenly included on a live webpage, damaging the platform's reputation.
Regarding the “AI label” disclosure that players have been asking about, VRChat believes such disclosures could help players quickly identify which projects used AI, but its internal team has yet to reach a clear consensus. VRChat is therefore open to introducing an AI labeling system but has not committed to implementing one.
In addition to its AI policy, VRChat plans to add a more convenient way to provide feedback on content within the game, allowing users to directly say whether they like or dislike particular content. This would help the company identify low-quality content and gradually improve its recommendation system.