Character.AI to ban teens from talking to its chatbots (www.engadget.com)

Character.AI announced it will stop allowing users under 18 to engage in open-ended, back-and-forth chats with its chatbots, effective November 25. Until then the company is rolling out a restricted “under-18” experience that curbs interactions to two hours per day (with further reductions before the deadline) and nudges teens toward creative use cases like role-playing, video or stream generation rather than companionship. The company also introduced an internally built age-assurance tool to tailor experiences by age and launched an “AI Safety Lab” to collaborate with researchers and industry on safety measures. CEO Karandeep Anand framed the move as a strategic pivot from AI companion to a creation-focused role-playing platform. The change is significant for AI/ML because it signals tighter operational and safety guardrails in response to regulatory scrutiny and real-world harms. Character.AI’s measures follow FTC inquiries and legal pressure—alongside scrutiny from state attorneys general and high-profile lawsuits alleging chatbots enabled self-harm—and reflect an industry trend toward stricter youth protections, verification, and content moderation. Technically, the shift raises questions about age-assurance methods, personalization constraints, and how reduced youth access will affect training/behavioral fine-tuning, logging and evaluation data. The Safety Lab could accelerate shared best practices (filtering strategies, RLHF safety tuning, evaluation metrics) but also highlights the broader trade-offs between model openness, user safety, and product utility.