Lab publication Credible

Taking Bold Steps to Keep Teen Users Safe on Character.AI

A Character.AI company blog post announcing the removal of open-ended chat for users under 18, new in-house and third-party age-assurance technology, and the founding of an independent nonprofit, the AI Safety Lab, dedicated to safety research for AI entertainment/companion features. A follow-up post on November 21, 2025 details the phased rollout, including new crisis-resource partnerships with Koko (emotional-support tools, with planned integration to identify high-risk content directly in-product) and ThroughLine (a verified helpline network spanning roughly 1,500 services across 170 countries).

Read at blog.character.ai Archived copy

Publisher

Character.AI

Published

29 Oct 2025

Added

2 weeks ago

DOI

—

Key Findings

Open-ended chat with AI Characters for users under 18 was removed in a phased rollout beginning November 24, 2025, preceded by a chat-time limit that was reduced from two hours to one hour per day for US teen users
Character.AI built an in-house age-assurance model, combined with third-party tools including Persona
Character.AI established and funded the AI Safety Lab, an independent nonprofit focused on safety research for AI entertainment/companion features, inviting other companies, academics, and policymakers to participate
A new partnership with Koko (free self-guided emotional-support tools for young people) is intended to eventually identify high-risk content directly within the product; a separate partnership with ThroughLine integrates a verified helpline network of roughly 1,500 services across 170 countries into Character.AI's under-18 off-boarding experience

Methodology Notes

Company announcement and follow-up rollout-detail blog posts; no accompanying technical safety evaluation or benchmark is presented. The AI Safety Lab nonprofit was newly founded at time of publication and had not yet produced research output as of this writing.