OpenAI announced Wednesday that its recently launched ChatGPT for Teens is being used modestly, with most interactions centered on school‑related learning. The company said the average teenager spends less than 15 minutes per day on the service and that built‑in reminders to pause usage appear to be effective. At the same time, a new study released by the nonprofit Common Sense Media concluded that the platform still poses an “unacceptable risk” for users aged 13‑17 because key safety mechanisms frequently fail.
OpenAI’s teen‑focused safeguards
When OpenAI introduced ChatGPT for Teens in August, it bundled a suite of protections aimed at reducing emotional dependence and limiting exposure to harmful content. The safeguards include blocks on romantic or sexualized language, prevention of the model from implying consciousness, and filters that stop sexualized imagery. Parents who link their accounts to a teen’s profile are supposed to receive alerts if the conversation involves self‑harm, suicidal thoughts, or disordered eating. Additional features provide reminders when a teen shares a “sensitive” image and automatic prompts to take a break after extended sessions.
OpenAI reported that these controls have led to an increase in learning‑oriented use, such as homework help and study advice. The firm noted that when a teen uses the chatbot for three consecutive hours or more, more than 80 % of the prompts still relate to educational topics. Break reminders appear to be impactful: nearly half of the users who receive a reminder stop interacting with the model shortly thereafter.
Independent study flags gaps
Common Sense Media’s research, conducted with multiple accounts registered as teens, found that several of the promised safeguards do not work reliably. The study observed that even after an hour of conversation about suicidal ideation, self‑harm, or disordered eating on newly created, parent‑linked accounts, no alerts were sent to parents. Alerts were only triggered for older accounts that had accumulated weeks of conversation history on sensitive topics.
The report also highlighted that the chatbot often failed to recommend crisis hotlines or professional help in these high‑risk discussions. While the model successfully avoided “sexual roleplay” with teen users, it continued to anthropomorphize, speaking in a friendly tone and sometimes presenting itself as a companion—behaviors OpenAI had pledged to curb. Tom Siegel, who leads the Youth AI Safety Institute at Common Sense Media, warned that the current guardrails could give parents a false sense of security.
Company response and broader concerns
OpenAI’s spokesperson responded by questioning the methodology of the Common Sense Media study, noting that much of the testing may have begun before parental controls were fully activated. The company said it welcomes rigorous independent evaluation but does not believe the findings accurately reflect how its teen safeguards function in practice.
The debate occurs amid growing scrutiny of how artificial‑intelligence tools affect young people. Incidents such as the February mass shooting in Tumbler Ridge, Canada, carried out by an 18‑year‑old who had discussed gun violence with ChatGPT, have intensified calls for tighter regulation and clearer accountability. Legal pressure on OpenAI regarding youth usage has also risen, prompting the firm to emphasize its ongoing development of safety features.
Both OpenAI and external observers agree that protecting minors in the age of conversational AI remains a work in progress. While the company points to limited, learning‑focused usage patterns as evidence of responsible adoption, the Common Sense Media study underscores persistent gaps that could expose teens to harmful content. The contrasting assessments suggest that further independent testing and possibly stricter policy measures may be needed before the platform can be deemed safe for under‑18 users.
Mitchell Landsberg is a Senior Technology Correspondent at News Raise. He covers consumer electronics, artificial intelligence, software developments, and digital privacy trends.
