PolyBuzz is an AI character chat app. Users pick an AI personality from a library that other users have built, and hold an open-ended conversation with it by text or voice. It is rated 18+ on both app stores, its own terms of use ban anyone under 18 anywhere in the world, and it has been installed more than 50 million times on Android alone.
It is one of the largest apps of its kind and almost nothing has been written about it. It launched as Poly.AI and renamed itself PolyBuzz. Google Play shows 50 million-plus downloads and 1.17 million reviews; the App Store shows 481,000 ratings. Character.AI has had years of newspaper coverage, lawsuits and attorney general attention. PolyBuzz, at this size, has had almost none, which is why most parents have never heard the name.
What a child actually does on it. You choose a character and talk to it. The characters are made by other users, given a personality, a backstory and a voice, and the conversation carries on for as long as the user wants it to. The company describes chatbots that “think and reply like real characters, complete with authentic voices”. There is no task and no end point. It is a relationship, delivered in an app.
Apple rates it 18+ and says why. Google rates it 18+ and does not. Apple lists sexual content or nudity, mature or suggestive themes, and cartoon or fantasy violence as frequent or intense, with unrestricted web access and user-generated content also flagged. Google Play gives the same 18+ rating with two stated reasons: “Users Interact, In-App Purchases”. A parent reading the Android listing is told nothing at all about sexual content.
Its terms ban under-18s. Its help pages advertise a mode for children. The terms of use, effective 1 June 2026, say: “You must have reached 18-year-old to use PolyBuzz and/or our Services”, and that the service is “strictly prohibit for minors (less than 18-year-old) worldwide”. The FAQ on the same site says the app offers “graded modes for different age groups” including a “Teen Mode specifically for children”. Both statements are published by the same company. They cannot both be the policy.
The age check is a date typed at sign-up, except in the UK. New users enter a date of birth. UK users are additionally sent to third-party verification providers, Ondato and Yoti, which is what the Online Safety Act requires. Everywhere else, the 18+ rule is enforced by asking.
The chats are private, including from the company. The FAQ says conversations are “completely private” and that “neither the creators nor Poly have access to the chat content between users and the AI characters”. Human moderation is described as applying to publicly recommended content, not to private chats. If a conversation goes somewhere it should not, there is nothing watching it and nothing for a parent to read afterwards.
Who runs it is not a single answer. The App Store names the seller as Cloud Whale Interactive Technology LLC. The terms of use on polybuzz.ai name Singularity Mutual Entertainment Culture Media Limited, a Hong Kong company, and set Delaware as the governing law. Subscriptions run from $9.90 to $29.90 a month, with coin packs from $2.49.
What this means for you right now
Look for the name, in both spellings. The app appears as PolyBuzz and was previously Poly.AI, so an older install may still carry the old name on the home screen.
The one control that works here is the store, not the app. PolyBuzz is rated 18+ by Apple and by Google, so setting a maximum age rating for downloads on the device blocks it from being installed at all. That is worth doing before a conversation about it, because the app itself offers a parent nothing: no linked account, no activity summary, no way to see what was said. Its own help pages describe no parental controls of any kind.
And treat the 18+ rating as the real information. An app that tells its own users they must be 18, and tells parents it has a mode for children, has not decided which of those is true. The rating is the part both app stores agree on.
App store listings, terms of use and FAQ checked 6 October 2026. Terms of use effective 1 June 2026.
Sources:
