OpenAI on the 19th released a "ChatGPT for teens" mode with features that can help learning while reducing the risk of teenagers becoming dependent on artificial intelligence (AI).
If a user enters their age as 13 to 17, or if the system estimates the user is under 18, the account is automatically switched to ChatGPT for teens mode.
The teens mode applies safeguards that strictly control output when teenagers ask questions or try to talk about self-harm, violence, eating disorders, risky behavior, or sexual or explicit content.
It also includes a feature that warns users when they upload sensitive or private photos. After a certain period of use, it provides reminders to take a break and continuously informs users that it is AI so they do not mistake ChatGPT for a person.
It is also designed not to express romantic feelings toward teenagers or encourage emotional dependence, and not to speak as if it has emotions or consciousness.
For teen accounts linked to parents, they can set "Quiet hours," which completely blocks their child's use of ChatGPT, or "Study hours," which sets the learning mode as the default.
The learning mode adds measures to prevent cheating in which teenagers simply copy answers provided by ChatGPT when doing homework. Even if a user asks a question, instead of providing an instant summary or answer, it throws questions back or walks through step-by-step principles to help them find a solution.
OpenAI said it designed the learning mode based on research by child and education experts.
Observers see OpenAI's launch of a teen chatbot with extensive added safeguards as an effort to resolve legal risk ahead of an initial public offering (IPO).
Recently, OpenAI has faced several related lawsuits as cases emerged in which teenagers who interacted with ChatGPT engaged in self-harm or made extreme choices, or used ChatGPT to plot crimes such as mass shootings or killings.