Ninety-five per cent. That is the share of the 21,595 respondents to VRChat’s community survey who said they wanted some kind of label for AI-generated content. The platform’s answer, published in an official blog post on 28 October, was no: no “AI-generated” tag, no AI filters. The reaction has been swift and furious.
VRChat — the free social hub where players build worlds, design avatars and gather in virtual space — has watched generative AI content proliferate across its platform, much as it has across the streaming services now facing class-action fights over AI. Fans had been begging for some way to tell human-made worlds from machine-made ones. The company asked them exactly how that should work — and then declined to follow the answer.
What the survey actually found
The scale of the response was itself remarkable: 21,595 people filled out the survey in a single week. Alongside the numbers, VRChat ran one-to-one interviews with creators — several of its own team members build content in their spare time — and found a community that wants transparency but cannot agree on what transparency means.
Asked how AI usage should be labelled across eight example categories, respondents produced 3,141 different combinations, with no single option chosen by more than ten per cent. Wanting a label, the data suggests, is not the same as avoiding everything that would carry one: only a third (33%) said they would deliberately skip an “AI-assisted” world, and 37% of those who wrote in offered some version of “AI as a tool is fine, as long as a person is doing the creative work.”
Why VRChat says no to an AI-generated tag
“While we think some creators would honestly label their content, we think others would try to pass off generative or assisted work as completely authored by hand,” the post reads. The team argues a label would be easy to dodge and could fuel “sleuthing” behaviour — users sniffing out suspected AI content and harassing creators, including those who use no AI tools at all. “We think this potentially creates negative behavior spirals we don’t want to encourage.”
“Harassing someone over the tools they use, or don’t use, is already harassment. We will protect creators from these types of behaviors.”
Key point: VRChat is not adding an “AI-generated” label or filters. A “Human-authored” mark and a “Tools used” field are being considered — but neither is scheduled.
There is a protective dimension to the refusal. Long-tenured creators who are experimenting with AI told VRChat they are afraid to show their work, fearing that disclosure would invite attacks. Others take visible pride in rejecting AI tools entirely, and the platform is wary of a label system that turns either stance into a target.
Review bombs and cancelled subscriptions
Players were not convinced. On Steam, VRChat picked up 178 negative reviews against 79 positive ones in the three days after the announcement — a 69% negative rate, against roughly 29% in the preceding weeks. On 29 October alone, 105 negative reviews landed against just 15 positive ones.
Many come from the platform’s most devoted users, including reviewers with more than 3,000 logged hours calling the decision a betrayal of its community of creators. The complaints cluster around process as much as policy: fans were asked for feedback, gave a near-unanimous answer, and watched it get passed over. “Then what was the point of the survey?” became a refrain across X and the official forums, while some reviewers say they have cancelled their VRC+ subscriptions and are urging others to do the same.
What is changing instead
VRChat says it will act where the community did agree. Misleading thumbnails are now explicitly against the Creator Guidelines, with offenders already contacted and an SDK reminder coming when creators pick world thumbnails. World likes and dislikes are targeted for the end of 2026 or early 2027, multi-image world galleries are planned for the first half of 2027, and “quality” is being made a core discovery metric.
The one label still alive is inverted: a “Human-authored” mark rather than an AI tag. VRChat leans towards it because, as the post concedes, “saying you used AI can feel like a confession.” But nothing is scheduled, and the definition is thorny — does human-authored mean no AI tools at all, or simply that a human made the creative decisions? If it ships, it would be optional, appear only on world detail pages, never affect discovery ranking, and a missing mark could never be treated as an accusation.
That leaves the rift the decision was meant to avoid open anyway. Whether a platform can hold together creators who see AI as a threat and creators who see it as a ladder is a question VRChat must now answer without the one tool 95% of its community asked for.
