ChatGPT for Teens Adds New Safeguards — but Safety Gaps Remain

ChatGPT now has a dedicated experience for teens, but its new protections are not a guarantee. OpenAI began a global rollout of ChatGPT for Teens on Aug. 18, adding stronger safeguards for eligible users ages 13 to 17 and optional controls for linked parents and guardians.

The teen experience can be enabled when account-provided age information, verified age, or OpenAI’s age-prediction system indicates a user is under 18. Availability is still expanding, and the company acknowledges that its automated systems can make mistakes, giving families more tools without providing complete oversight.

What ChatGPT for Teens changes

Under the new ChatGPT for Teens experience, eligible accounts receive stronger protections by default. OpenAI says safeguards more cautiously handle graphic violence, risky viral challenges, sexual or romantic roleplay, extreme beauty standards, and unhealthy dieting, though they cannot guarantee every response will be appropriate.

The rollout comes amid broader pressure on technology companies to protect younger users. In August, a New Mexico court ordered Meta to overhaul teen safeguards on Facebook and Instagram, including protections involving AI chatbots.

Parental controls are a separate layer. Once accounts are linked, a parent or guardian can manage quiet hours, Voice Mode, image generation, saved memories, Study Mode, and whether conversations may be used to improve OpenAI’s models. Parents cannot read or monitor a teen’s chats through those controls.

The rollout follows a November 2025 evaluation by the Family Online Safety Institute of OpenAI’s earlier parental controls. Using a simulated teen account, FOSI found weaknesses involving age assurance, delayed enforcement, risky-content exposure, and in-chat purchases.

That testing predates ChatGPT for Teens and does not establish how the current product performs. OpenAI’s current documentation nevertheless cautions that parental-control updates may not take effect immediately and that some changes apply only to new conversations.

Where the safety gaps remain

OpenAI’s age-prediction system uses account signals including general topics discussed, time-of-day usage, how the account is used, and how long it has existed. The company acknowledges that the system can make mistakes, underscoring a broader industry challenge as platforms experiment with privacy-focused age signals rather than relying solely on self-reported birth dates.

Age assurance also becomes an enforcement issue once access depends on getting the classification right. Meta, for example, recently blocked more than 750,000 under-16 accounts in Australia, showing the scale at which age-detection systems can affect access when platforms must enforce minimum-age rules.

A 2025 Common Sense Media survey found that 33% of teen AI-companion users had at least once chosen an AI companion over a real person for an important or serious conversation. ChatGPT is a general-purpose chatbot rather than a dedicated companion service, so that figure is not a measure of ChatGPT use specifically.

The American Psychological Association advises parents not to treat general-purpose generative AI chatbots as substitutes for professional mental health care. Its guidance recommends discussing AI use with teens, setting boundaries together, and taking concerns involving self-harm, suicide, or a mental health crisis to qualified human support.

ChatGPT for Teens adds meaningful safeguards but does not create a monitored or error-proof environment. Parents cannot see conversations through the controls, age prediction can be wrong, and automated content protections cannot ensure every response will be appropriate.

Read more: The challenge extends beyond AI chatbots, as Australia’s teen social media rules expose the limits of age verification across major online platforms.