Is PolyBuzz AI safe for kids?

PolyBuzz is a character-roleplay app in the same broad category as Character.AI and Janitor AI. Users browse thousands of user-created "bots" — fictional people, anime characters, celebrities, original roleplay scenarios — and chat with them one-on-one. The product is optimized for long, immersive, emotionally-engaging conversations.

That design goal is the source of nearly every safety concern below. Engagement, not child development, is the metric the platform is tuned for.

• Romance and dating-style characters dominate the front page. Many bots are explicitly framed as flirtatious partners, even when not labeled NSFW.
• Age verification is self-declared. A child can tap through to access the platform without meaningful gating.
• There is no parent dashboard, no transcript export, no safety-event notifications, and no way for a guardian to set topic boundaries.
• User-generated characters mean the safety floor depends on whatever the bot's author wrote. Filters can be bypassed by roleplay framing — a documented pattern across every character-chat app of this type.
• Conversations are private to the user account. A parent cannot read what their child has been saying or what the bots said back.

The lessons from the Character.AI and ChatGPT teen-harm lawsuits over the last 18 months are consistent. A platform a child uses unsupervised needs, at a minimum:

• An age-appropriate content model — not a general-purpose model with filters bolted on.
• Identity protection so the AI never sees the child's real name.
• Parent-set topic boundaries that the model actually respects.
• Full transcripts available to the parent, on demand.
• Real-time safety-event alerts when something concerning is said.
• Time and usage limits the child cannot bypass.
• A business model that does not depend on maximizing time-on-app.

PolyBuzz does not offer these. That is not a knock on its team — it is simply a product built for a different audience.

We built Sorin because the choice between "no AI" and "the same tools adults use" is unacceptable for children growing up into an AI-native workforce. Sorin is a child-first AI companion with:

• Reading level, tone, and topic scope that adjust to the child's age (6–15).
• Name swapping before any message reaches the model — the AI never sees your child's identity.
• A Socratic teaching style: Sorin asks questions and guides discovery rather than handing back answers.
• An AI Policy editor where you set allowed topics, off-limits subjects, and homework rules per child.
• Full transcripts, weekly plain-English summaries, and real-time safety events in the parent dashboard.
• Daily and weekly time limits the child cannot override.
• A subscription model — no ads, no engagement-maximizing incentive.

For the full security and oversight breakdown, see our Security & Privacy page.

For an adult or older teen who understands roleplay-chat platforms, PolyBuzz behaves about how you'd expect a Character.AI-style product to behave. For an elementary, middle, or early high-school child, the honest answer is no. There is no parental oversight layer, the default content surface skews toward romance and dating roleplay, and the safety model relies on filters that user-generated content consistently finds ways around.

If you want your child to grow up fluent with AI without exposing them to a product built for adults, that's exactly what Sorin is designed for.