Common Sense Media labels ChatGPT for Teens an unacceptable risk despite new features
AI Summary
New research claims OpenAI's safety guardrails for teenagers are unreliable, even as the company rolls out new college planning and study tools.
Non-profit Common Sense Media has deemed OpenAI's ChatGPT for Teens an "unacceptable risk," reporting that critical safety guardrails and parental alerts for mental health crises frequently fail.
Common Sense Media released research on Wednesday calling for a ban on users under 18 until OpenAI's teen-specific guardrails are proven reliable. The group's report found that while some protections against sexual roleplay worked, other features failed to deliver on commitments. The researchers noted that engagement cues remained pervasive even when users were in crisis, leading the non-profit to give the tool a failing score for three out of five "Red Line" severe harms.
OpenAI disputed the findings, claiming that Common Sense Media’s testing methodology may have concluded before parental controls were fully activated. An OpenAI spokesperson stated that the study does not accurately reflect how the safeguards work in practice. To counter concerns about addictive behavior, OpenAI reported that the average teen spends less than 15 minutes a day on the platform and that 80% of long-session prompts are related to learning.
Amid the safety criticism, The Verge reported that OpenAI is introducing a new College Planner tool to the teen mode. The feature is designed to help students in grades 10 through 12 track application deadlines, financial aid steps, and tasks for multiple schools. Other upcoming updates include a multi-photo capture tool for creating PDFs and the ability to generate flashcards from student notes.
TechCrunch noted that the scrutiny comes as lawmakers target chatbot designs that maximize user attention. While OpenAI claims features like break reminders successfully prompt half of teen users to stop after a warning, Common Sense Media maintains that the tool still assists with homework and fails to notify parents during discussions of self-harm or eating disorders as promised.
Background
OpenAI launched ChatGPT for Teens in August with features intended to prevent emotional dependence, block sexualized imagery, and provide parental notifications for problematic usage.
How outlets covered it
Get our daily briefing first.
We're starting a short daily email with the stories most outlets are covering. Join the list to get the first one.
Free. No spam.
Discussion
No comments yet. Be the first to start the conversation!