VRChat won't label AI-generated content, and Steam reviews turn against it

VRChat says an AI label would invite dishonest tagging and harassment. Most of its surveyed users wanted one anyway, and its Steam reviews show it.

Mason Reed

Abstract illustration of glowing floating islands in a twilight sky, some hand-carved and some uniform, with blank luminous tags drifting between them

VRChat has decided not to label AI-generated content, even though its own survey found most respondents wanted some kind of label, and Steam users have answered with a wave of negative reviews.

VRChat's original announcement couldn't be checked directly against a registered first-party source for this article. Everything below is as reported by Engadget, Automaton and Road to VR, credited to each.

What VRChat decided

Automaton reports that on October 29, VRChat set out its plans following a user survey about AI on the social VR platform. Engadget describes the announcement as a support post explaining the survey's results. Engadget reports that 95 percent of respondents said they wanted some form of label to designate AI usage, but the company said it will not add a tag for AI-generated content, or a filter that removes it from search results.

Engadget quotes the company's reasoning: "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 went on, in the same quote, to warn about "'sleuthing' behavior, where folks would attempt to sniff out 'AI' content or might harass other creators, even if they are genuinely not using any AI tools in their workflow." Road to VR quotes the next line too: "We think this potentially creates negative behavior spirals we don't want to encourage."

The door isn't fully closed. Engadget reports that VRChat is still weighing a label for "Human-authored" content and a "Tools used" field, without a definite position yet. Road to VR quotes the company saying that if it adds the mark, "it'd be optional, it'd only show on the world's detail page, it wouldn't affect discovery ranking, and a missing mark would never be an accusation or something you can report."

What the survey actually said

The breakdown Automaton reports suggests the picture behind the 95 percent figure is more divided. According to Automaton, only 33 percent of respondents said they wanted to avoid worlds made with AI, while 37 percent said AI is just a tool and is fine as long as a person does the creative work. Those groups can overlap with the people who asked for labels. Across eight example categories of AI use, Automaton reports, answers split into 3,141 different labelling combinations, and no single combination was backed by more than 10 percent of respondents. Road to VR puts the number of people surveyed at more than 21,000, and notes the results are no longer viewable online.

What VRChat is doing instead

Road to VR reports that VRChat will ban what it calls "AI-generated inaccurate thumbnails" used to advertise worlds, and add ways to like, dislike and give feedback on worlds inside VRChat. Automaton adds the timing: the like and dislike vote is aimed at the end of 2026 and due by early 2027 at the latest, and world listings are planned to show several images instead of a single thumbnail in the first half of 2027. Automaton also reports that VRChat contacted creators whose worlds broke its updated guidelines and asked them to fix them.

Engadget notes that the changes VRChat is making are the ones it said had "alignment across the community," including accurate thumbnails, better discovery tools and "quality as a core metric."

The Steam backlash

Users weren't won over. Automaton reports that about 950 negative Steam reviews were posted on October 29 alone, and that VRChat's reviews over the past 30 days stood at 33 percent positive out of roughly 4,000. Road to VR reports the same 33 percent figure, calls the 30-day rating "Mostly Negative," and says the surge pushed VRChat's overall lifetime rating from "Mostly Positive" to "Mixed."

Automaton also spotted an "Artificial Intelligence" tag appearing on VRChat's Steam store page on November 1 and disappearing the next morning, Japan time, according to SteamDB records. Steam users as well as developers can apply tags, and Automaton says it can't say for sure the tag is connected, but suggests many users may have added it in protest. As of Automaton's report, VRChat had not made any further statement on its forum or its official X account.

For scale, Road to VR points back to July 2022, when VRChat's decision to add anti-cheat software drew more than 33,000 negative reviews, according to SteamDB.

Mason's take: VRChat's worry about witch hunts isn't silly; a label nobody can verify mostly rewards whoever lies best. But the survey numbers also explain the anger. When 95 percent of survey respondents ask for something, a firm no with a maybe-later optional badge sounds like a shrug, even when the reasoning behind it holds up. The thumbs-up and thumbs-down vote is the real test. If it arrives on time and actually buries misleading worlds, the label fight may matter less than it looks this week. If it slips, the review score will keep doing the talking.

Sources

More from Xarmo News