ChatGPT teen account safety features are problematic, new report finds

Key Points:

• Researchers at Common Sense Media’s Youth AI Safety Institute tested ChatGPT’s new teen safety features using fictional teen personas and found that key safeguards, including parental notifications for crisis situations, often failed to activate as intended.
• Despite OpenAI’s rollout of ChatGPT for Teens with built-in protections and parental alerts, Common Sense Media’s report revealed that hundreds of prompts involving self-harm and suicidal ideation did not trigger any parental notifications, raising concerns about the system’s reliability.
• OpenAI disputed the report’s findings, citing methodological flaws and stating that parental controls require linked parent-teen accounts to be active for several hours before alerts function properly; however, Common Sense Media stands by its conclusions.
• The report also highlighted that ChatGPT often uses friend-like, empathetic language in mental health conversations, which may encourage teens to treat the AI as a substitute for human support, a trend experts warn can be harmful and delay seeking real help.
• Mental health professionals emphasize that while AI can feel responsive, it cannot replace human relationships or emergency interventions, and they caution that adolescents are particularly vulnerable to forming unhealthy attachments to AI chatbots during critical developmental stages.