OpenAI launches a version of ChatGPT for teens

OpenAI has introduced a new version of its flagship chatbot ChatGPT designed exclusively for teenagers, with stricter security and monitoring measures, after facing criticism and legal complaints that it pushed some users to suicide. Dubbed “ChatGPT for Teens,” the version with modified parameters automatically activates for any user who registers claiming an age between 13 and 17, or who the company detects as a minor through usage patterns or account seniority.

Most of the security measures used in the ChatGPT for Teens model were already in place for ChatGPT accounts belonging to minors, such as restrictions on content related to self-harm, violence, eating disorders, dangerous activities, romantic or sexual conversations, and explicit sexuality.

Hours of silence and ‘Study Mode’ for teenagers

The program also offers more frequent break reminders, which kick in after 90 minutes of continuous activity in a three-hour window, “quiet hours” with service blackout, and Study Mode, which responds to users with leading questions rather than providing immediate direct answers,”to help teens understand the material“, according to OpenAI.

Next to the hours of silence, the company also introduces “study hours”a time slot in which Studio mode remains active by default. The latter includes quizzes and educational visualizations designed to help teenagers practice and verify what they have learned, and is able to recognize when a child is trying to circumvent a task, returning him to guided mode.

Other restrictions and parental controls

Other restrictions in the ChatGPT for Teens model include preventing the chatbot from implying that it has personal feelings towards the user or has its own conscience, and some modes such as voice mode are disabled. Parents are offered the opportunity to link their accounts with those of their teenage children, and to be notified if certain high-risk situations, such as self-harm or suicide, are discussed.

Parental advisories are reviewed by human moderators, and specific details of conversation logs are removed to protect privacy. Regarding eating disorders, the company explained that it has added dedicated notifications, limiting the information shared and focusing on the moments when offline support can make a difference.

The legal and regulatory pressure on ChatGPT

The new modality for teenagers comes against a backdrop of growing legal and regulatory pressure. OpenAI faces several wrongful death lawsuits, including from the parents of Adam Raine, 16, a teenager who allegedly took his own life using methods suggested by ChatGPT. The US Federal Trade Commission has also launched an investigation into AI-powered chatbots that serve as digital companions, focusing on potential risks to children and teenagers.

Like social media companies, AI giants also face growing regulatory pressure: Congress has introduced several bills to regulate chatbots, and California requires security measures for assistants who serve as companions or confidants. Other AI companies have also received criticism for issues related to use by minors, with some, like the rival Anthropicwho have chosen to limit the use of their model to adults. OpenAI also noted that nearly a third of American teenagers use AI-powered chatbots every day, according to a Pew Research Center study published in December.