VRChat Declines AI Labels Despite 95% Survey Demand
VRChat will not add tags for AI-generated content at this time, the company said on October 29, 2026, even though 95 percent of respondents to its user survey said they want some form of label.
Engadget reported that VRChat will add neither a tag for labeling AI-generated content nor a tag for filtering such content out of search results. The company said it will keep weighing two possible additions, a "Human-authored" label and a "Tools used" field, but has not taken a definitive position on either.
The reasoning centers on how labels would behave in practice. VRChat said some creators would label their content honestly, while others would try to pass off generative work, meaning content produced by AI models, or AI-assisted work as entirely hand-made. It also said AI labels could produce "sleuthing," in which users try to detect AI content or harass creators, even creators who do not use AI tools.
The survey produced little consensus beyond the demand for a label. According to VRChat, apart from wanting some kind of label for AI usage, there was very little agreement among respondents. That finding was laid out in an Official Notes post titled "Where we're landing on AI in VRChat," published October 28, 2026, on the VRChat forums.
The underlying questions predate the survey. An earlier Official Notes post, "Regarding AI Usage on VRChat," published October 6, 2026, asked what AI labels should be, what they would mean, whether AI usage can be detected, and whether authors can be trusted to label their own content. Each of those questions bears directly on the options VRChat is still weighing.
Rather than labeling, VRChat said it would adopt changes including accurate thumbnails, better discovery tools and "quality as a core metric." The announced changes address how content is presented and found, not how it was produced.
In my view, the central difficulty is that a provenance label, a record of how a piece of work was made, is only as reliable as the person applying it. A platform has no dependable way to verify that answer after the fact. VRChat's own questions about detection and self-labeling point at that gap. The company's stated alternative sidesteps it by judging the work rather than the toolchain behind it. That approach is coherent, but it asks more of discovery systems and moderation than a checkbox does, and it will only be tested once the changes are in use.
The broader context here is a platform built around user-made avatars, worlds and experiences, where creators and viewers both have reason to want an honest signal about what they are looking at. Whether quality-based discovery can supply that signal without the sleuthing VRChat described is an open question. If it can, the upside is a system that rewards craft and presentation over the simple fact of which tools were used.
