1 hr ago
OpenAI Reports Teen ChatGPT Use as Safety Concerns Grow
OpenAI shared information about how teenagers use ChatGPT.
It said teens use it for less than 15 minutes a day on average.
The company also said very few teens chat for more than three hours in a row.
OpenAI says reminders help some users take a break.
It has a teen mode with parental controls and limits on sensitive topics.
But Common Sense Media tested accounts and said safety alerts for serious topics were sometimes late or missing.
It also said the chatbot did not reliably point users to crisis help.
OpenAI said the tests may not show how the safeguards work once fully activated.
The disagreement comes as people worry about how AI chatbots may affect young users.
OpenAI said teens spend an average of under 15 minutes a day on ChatGPT, and fewer than 2% use it for more than three consecutive hours.
The company said nearly half of teens who saw a break reminder took a break or ended the chat within five minutes.
OpenAI said its teen mode, rolled out in August, includes parental controls and limits conversations on sensitive topics.
Common Sense Media said its tests found delayed or missing parent alerts and unreliable suggestions of crisis hotlines or professional help.
OpenAI disputed whether the tests reflected its safeguards, saying parental controls can take hours to activate after accounts are linked.
- Who
- OpenAI, teenage ChatGPT users, and Common Sense Media.
- What
- OpenAI released figures on teen ChatGPT use, while Common Sense Media raised concerns about teen safety alerts and crisis guidance.
- Where
- The articles do not specify a location.
- When
- OpenAI released its report on Wednesday; ChatGPT for Teens began rolling out in August.
- Why
- The report and testing address concerns about AI chatbots' risks to adolescent mental health and safety.
OpenAI's account
Common Sense Media's findings
Teen safeguards and alerts
OpenAI's account
OpenAI said its teen safeguards include parental controls and sensitive-topic limits, and that test results may not reflect how safeguards work after activation.
Common Sense Media's findings
Common Sense Media said its test accounts did not produce timely alerts about conversations involving suicide, self-harm, or eating disorders, and that crisis referrals were unreliable.
How to interpret account testing
OpenAI's account
OpenAI said linking parent and teen accounts and activating alerts may take hours, so some tests may have begun or ended before controls were fully active.
Common Sense Media's findings
Common Sense Media said some test accounts had been linked long enough and some had not, but none produced timely alerts; notifications appeared only after weeks of sensitive-topic history.
Key facts
- Average teen use
- Under 15 minutes per day, according to OpenAI.
- Extended use
- Less than 2% of teens use ChatGPT for over three straight hours, according to OpenAI.
- Break reminders
- In nearly half of conversations where a reminder appeared, the user took a break or ended the chat within five minutes, OpenAI said.
- Teen mode
- OpenAI began rolling out ChatGPT for Teens in August; it can enable parental controls and limit sensitive-topic conversations.
- Common Sense Media testing
- The nonprofit tested more than a dozen accounts and reported that alerts were not reliably timely and crisis guidance was unreliable.
- OpenAI response
- The company said account linking and activation of safety alerts can take hours, and argued that testing may not reflect fully active controls.
- Planned study tools
- OpenAI is considering a college application planning tool, flashcards, and easier quiz-making tools.
Quotes
OpenAI spokesperson
A spokesperson for OpenAI, the company that developed ChatGPT.
“The bulk of their testing may have begun and concluded before activation of parental controls was complete.”
CNBC TV 18
Sources
Related news
Florida Seeks Court Freeze on OpenAI Model Development
Gen Z’s Growing Friendship With AI Shows Promise And Risks
WhatsApp Rolls Out Optional Parental Controls for Teenagers
OpenAI Warns Over 100 Organisations About Rogue AI Activity
Former Google safety chief urges AI crash tests for children
Mother Tells Congress ChatGPT Advised Son on Deadly Drug Mix
ChatGPT Teen Safeguards Face Questions Among Indian Users
Google unveils Gemini and Search AI learning tools for students
