OpenAI launches ChatGPT for teens amid youth safety concerns

SINGAPORE - A new variant of ChatGPT designed for teenagers will be rolled out from Aug 19 amid global youth safety concerns.

ChatGPT for Teens, which will automatically apply to registered users under the age of 18, comes with reduced exposure to content that may be harmful or developmentally inappropriate, said OpenAI in a blog post on Aug 18.

This includes self-harm, violence, eating disorders, dangerous activities, and explicit sexual or graphic content.

“These stronger protections are designed to reinforce healthy, real-world relationships,” said OpenAI, adding that a full roll-out to all users around the world will take a few weeks.

ChatGPT can also be used without a registered account.

During account registration, users will be asked to declare their age. OpenAI’s own age prediction system will also estimate an account holder’s age through usage patterns, among other indicators.

ChatGPT for Teens will also include reminders that caution teens against uploading private or sensitive images. During account registrations, teens will be introduced to learning and safety features.

The safety features block ChatGPT from engaging in romantic or sexualised roleplay, encouraging emotional dependence, or implying that it has feelings or consciousness.

Parents will be able to link their accounts to their teen’s account, and receive safety notifications in high-risk situations, such as signs of eating disorders.

“Parents don’t have access to their teen’s conversations, except in rare cases where our system and trained reviewers detect possible signs of serious safety risk,” said OpenAI, adding that parents may be notified, but only with the information they need to support their child’s safety.

Other parental controls include preventing teens from viewing graphic content and viral challenges, and setting hours when ChatGPT cannot be used.

To link parent and teen accounts, teens must accept their parent’s invite on their own account.

ChatGPT for Teens also offers differentiated learning experiences, which includes Study Mode, which uses guiding questions and step-by-step support to help teenagers understand the material. Homework reminders will also redirect teenagers back to Study Mode when it recognises that the user is trying to get direct answers to a question.

Generative AI chatbots such as ChatGPT have become an indispensable tool for many, but have also courted controversy for generating harmful content linked to cases of suicide.

In 2025, parents of Adam Raine sued OpenAI over the death of their teenage son, alleging that ChatGPT had encouraged the 16-year-old to take his own life. The lawsuit detailed how Adam first began using ChatGPT for school work, but he later start discussing methods of suicide with the chatbot.

ChatGPT has also been used to generate sexualised images or graphic violence.

Concerns over generative AI’s ability to spew such content have also spurred other major players to continuously roll out safeguards that protect teenagers.

Teen accounts on Meta’s Instagram are made private by default, can only accept messages from accounts they follow or have connected with, and blur images in messages that contain nudity.

Instagram is integrated with Meta AI chatbot, which reflects a larger trend of social media platforms that now offer AI services. The main players include Facebook with Meta AI, and X with Grok.

In a statement on Aug 18, Meta said it is working on developing a feature that would notify parents if their teen attempts to engage in conversations related to suicide or self-harm with Meta AI.

Earlier in February, Minister of State for Digital Development and Information Rahayu Mahzam announced that the local authorities are studying the need for safegaurds to curb online harms perpetuated by AI chatbots.

During the debate on her ministry’s budget, Ms Rahayu also acknowledged the dangers of AI chatbots, especially in social media services, as children can access them more easily.

Existing measures to protect youths in Singapore from online harm include a range of age assurance measures that were rolled out by app stores in April, to control app downloads by those below 18. This is done to meet requirements in the Code of Practice for Online Safety for App Distribution Services enforced by the Infocomm Media Development Authority.

The authority has plans to extend age assurance requirements to major social media services.

While the government has not ruled out an age-based blanket ban on social media, an expert panel formed by the Ministry of Health discouraged this in a report submitted to the authorities earlier in July.

Instead, the eight-member expert panel recommended tackling specific features that could expose users to safety risks, prolong usage, or worsen mental health - such as video autoplay functions and algorithmically curated feeds.