Common Sense Media Finds ChatGPT for Teens Not Significantly Safer Than Previous Version
A new evaluation by nonprofit organization Common Sense Media has found OpenAI's ChatGPT for Teens, launched in August to provide safer AI access for users aged 13–17, does not significantly improve safety compared to the standard version. The Youth AI Safety Institute tested over 4,000 prompts and concluded that key features like parental alerts failed during crisis-related simulations involving suicide or self-harm. Additionally, ChatGPT's responses sometimes implied emotional awareness or consciousness—contrary to OpenAI’s stated goal of discouraging such perceptions. The review also found no reclassification when an adult account was used by someone under 18 and that safety filters did not activate in those cases. While some features like the refusal of explicit sexual roleplay worked as intended, many others fell short. Common Sense Media gave ChatGPT for Teens—and similar platforms such as Grok and Meta AI—an
Topics: AI safety Common Sense Media OpenAI teenagers