Why My Conversations with OpenAI's 'ChatGPT for Teens' Made Me Very Worried
- OpenAI's 'ChatGPT for Teens' fails to consistently trigger crisis resources for students mentioning suicide or self-harm.
- Parental notification features are essentially broken, failing to alert parents to sensitive or high-risk conversations.
- Study mode is a joke; students can bypass 'tutoring' features with a single click to get instant homework answers.
- Experts warn that we are playing with fire by allowing AI to build parasocial relationships with vulnerable youth.
Brief Summary
OpenAI’s push to market a safer version of ChatGPT for teenagers is being exposed as a hollow PR exercise. Despite claims of new 'guardrails' meant to protect minors from self-harm and academic dishonesty, testing by both the New York Times and Common Sense Media reveals that the bot remains a shortcut for cheating and a fickle companion for those in crisis. The 'study mode' meant to encourage learning is easily circumvented, and critical safety notifications to parents often fail to fire entirely.
Why This Matters
If you are a parent, you need to understand that the 'safety features' on your child’s devices are likely providing a false sense of security. You are essentially handing your teenager a direct line to a machine that doesn't reliably understand the difference between a student asking for help and a child in distress. Relying on these algorithms to police your child's mental health or study habits is a dangerous gamble, as the technology is currently prioritizing engagement over genuine safety and academic integrity.