90 Days Gen AI Risk Trial -Start Now
Book a demo
Chatbots·Free tier; c.ai+ $9.99/mo; no enterprise plan·character.ai

Character.AI

Character.AI lets users chat with user-created AI personas for roleplay, companionship, and creative writing, with a largely consumer, youth-skewing audience and minimal enterprise controls.

Risk Score
High
7/10

Independent assessment across data handling, compliance, security and transparency.

Overview

Character.AI is a consumer chatbot platform where anyone can create and share AI personas (real, fictional, or original) built on proprietary LLMs. Users hold long-form, memory-enabled conversations with these characters for entertainment, roleplay, emotional support, tutoring, and creative writing. The product skews heavily to teens and young adults, and has been the subject of multiple lawsuits and regulatory scrutiny related to minor safety, self-harm content, and parasocial dependency. The platform offers a free tier plus Character.AI Plus ($9.99/month) for faster responses and early features. Enterprise controls are essentially absent: no SSO, no admin console, no DPA for business use. Character personas are user-generated and not vendor-moderated at source, so behavior, claims, and persona 'expertise' are uncontrolled. Treat as a non-work-appropriate consumer app.

Risk factors

4
  • Consumer-focused product with no enterprise plan or SSO
  • User conversations used to improve models by default
  • Youth-skewing user base raises compliance concerns for employers
  • Minimal admin controls for organisation-wide visibility

Recommendations

3
  • Block domain at egress if enterprise policies prohibit consumer AI
  • Educate employees on not pasting proprietary data
  • Monitor DNS / browser telemetry for usage

Data handling

Storage
Conversations and user-created characters stored indefinitely on Character.AI servers (US-based, GCP). No customer-controlled regional residency.
Retention
Chats retained until the user deletes them or closes the account; deletion requests honored within ~30 days per privacy policy.
Training on inputs
User conversations and character interactions are used to train and improve Character.AI models by default, with no consumer opt-out.