A leading child safety watchdog group has concluded that ChatGPT for Teens, OpenAI’s AI tool designed for users aged 13 to 17, does not meet basic safety standards for underage users. The nonprofit Common Sense Media released its evaluation on Wednesday, assigning the platform an “unacceptable risk” rating after testing more than 4,000 prompts between July and September.
OpenAI launched ChatGPT for Teens in August, billing it as a safer version of its chatbot with enhanced controls for parents and built-in safeguards to limit exposure to harmful content. The company stated its goal was to enable teens to use AI “responsibly to learn, create and explore.” However, Common Sense Media’s findings suggest these protections are largely ineffective.
Key findings from the evaluation:
- Parental alerts for crisis-oriented prompts failed to trigger in simulated tests of self-harm, suicidal ideation, and mental health crises.
- Explicit sexual roleplay was blocked, but other safeguards did not perform as promised.
- The organization recommended that teens avoid using the platform until significant improvements are made.
OpenAI’s response to the report has not been publicly detailed as of this publication.
How the Safeguards Were Tested
Common Sense Media’s Youth AI Safety Institute conducted its evaluation by simulating conversations across multiple teen personas, including those in crisis. Researchers created more than a dozen accounts linked to parental controls before engaging with the chatbot. The tests revealed that while some safeguards, such as blocking role-playing, functioned as intended, most did not meet safety expectations.
Tom Siegel, executive director of the Youth AI Safety Institute, stated, “At this point, we’re recommending that teens don’t use it.” The organization’s concerns center on the chatbot’s failure to adequately protect vulnerable users from harmful content or provide timely interventions in crisis situations.
What OpenAI Promised vs. What Was Delivered
When OpenAI announced ChatGPT for Teens, it highlighted several features aimed at improving safety:
- Default “teen-specific mode” with stricter content filters.
- Parental controls to monitor usage and receive alerts.
- Refusal of explicit sexual roleplay and claims of sentience.
- Promotion of healthy use habits through guided interactions.
However, Common Sense Media’s testing found inconsistencies in enforcement. For example:
- Suicide and self-harm prompts did not trigger parental alerts, even after prolonged simulated conversations.
- Mental health-related queries (e.g., psychosis, eating disorders) were not consistently addressed with appropriate safeguards.
- Parental notifications failed to activate in multiple test scenarios.
Lauren Jonas, OpenAI’s head of youth and families, previously told NPR that the teen mode was designed to “enable teens to better use ChatGPT as a learning tool while limiting exposure to harmful and developmentally inappropriate content.” The company has not issued a public response to the watchdog group’s findings.
Broader Context: AI Safety for Minors
The controversy over ChatGPT for Teens reflects ongoing concerns about AI’s impact on young users. Common Sense Media’s report is part of a broader push for stricter oversight of AI tools targeting minors. Earlier this year, the organization also assigned an “unacceptable risk” rating to Meta AI and X’s Grok, citing similar failures in safety protections.
Robbie Torney, head of AI and digital assessments at Common Sense Media, emphasized the need for “stronger built-in safety protections” before such tools are deemed suitable for teens. “We found very little evidence to show that this new version of ChatGPT is safer than the previous version,” Torney stated.
What Parents and Educators Should Know
Common Sense Media has outlined key takeaways for adults considering the tool:
- Parental alerts are unreliable and may not activate during crises.
- Not all promised safeguards are fully functional, including mental health-related protections.
- Alternative learning tools may offer better safety controls for teens.
The organization has not ruled out revising its assessment if OpenAI implements significant improvements.