Interview on Stirile ProTV (September 2025)

In September 2025, I joined Stirile ProTV to discuss OpenAI’s new parental controls in ChatGPT for teen users — what they enable, why they matter, and how to deploy them responsibly.


We covered linking parent and teen accounts, setting age‑appropriate rules, managing memory and chat history, and proactive alerts when the system detects moments of acute distress — all designed to increase safety while preserving autonomy.

"They are just numbers that, during training, learned to speak the way humans speak. The system evolves to favor behaviors where it is always ready to help — it simulates empathy."

Stirile ProTV Interview — September 2025

We also discussed strengthening safeguards in long conversations, routing sensitive chats to higher‑reasoning models, and keeping humans in the loop for timely escalation.

Parental Controls and Teen Safety

Key capabilities introduced include:

  • Linked parent‑teen accounts for oversight
  • Age‑appropriate response rules and feature controls
  • Memory and chat‑history management
  • Alerts when “acute distress” is detected
  • Routing sensitive chats to higher‑reasoning models
  • Stronger safeguards for long, multi‑turn conversations

Together, these features aim to provide effective oversight without turning the chatbot into a substitute for real‑life support.

Understanding “Simulated Empathy”

Large models optimize for helpfulness and can sound empathetic, but they do not understand or feel. Clear guardrails ensure supportive responses that steer toward human connection and professional help when needed.

"Empathy‑like responses can support teens, but they must always point to human care, not replace it."

Practical Guidance for Families and Schools

Start with clear agreements, review settings together, discuss limits of the system, teach escalation to trusted adults and hotlines, and revisit controls as needs evolve.

This interview on Stirile ProTV highlighted a balanced approach to AI for teens: meaningful parental oversight, robust safety rails, and transparent communication — so support is available, and help is escalated when it really matters.