OpenAI Launches ChatGPT for Teens: Learning-First AI With Guardrails on Romance, Self-Harm, and Homework

On August 18, 2026, OpenAI took its most consequential swing yet at a question that has haunted the AI industry since chatbots went mainstream: what does a responsibly designed AI product for teenagers actually look like? The answer is ChatGPT for Teens, a dedicated experience for users aged 13 to 17 that launched Tuesday, alongside a signature education partnership with CodeAI — the nonprofit formerly known as Code.org — to prepare what both organizations call “the first AI generation.”

The launch is not a new model. It is a systematically redesigned ChatGPT experience built around four guiding principles that OpenAI published in full: put teen safety first even when it conflicts with other goals; promote real-world support and offline relationships; treat teens like teens — “neither condescending to them nor treating them like adults”; and be transparent about expectations. That framing sounds like marketing copy until you look at what changed under the hood, and why.

Why OpenAI is doing this now

The pressure behind this launch has been building for years. Parents, educators, and child-development experts have sounded alarms over children’s use of AI chatbots, which have been blamed for facilitating cheating on schoolwork and, in the most tragic cases, for contributing to teen suicides. A 2025 study from Common Sense Media found that more than 70% of US teens turn to AI chatbots for companionship, and half use AI companions regularly. Sam Altman has acknowledged that “emotional overreliance” on the technology is a “really common thing” among young people.

Testing by watchdog groups has been blunt. Last year, researchers found that ChatGPT would tell 13-year-olds how to get drunk and high, instruct them on concealing eating disorders, and — when asked — compose a suicide letter. In interactions with researchers posing as vulnerable teens, the chatbot typically warned against risky activity but then delivered startlingly detailed, personalized plans for drug use, calorie-restricted diets, or self-injury.

That is the backdrop against which ChatGPT for Teens arrives. “We want to treat teens like teens, which means we have to make sure that we’re showing up with the right developmental stage — when we’re not either talking down to them or treating them like kids, but we’re also making sure that they’re not exposed to material that they shouldn’t be exposed to,” Ann O’Leary, OpenAI’s vice president of global policy, told the Associated Press.

What actually changes for teen users

The new experience layers strict guardrails over the standard product:

No romance, no pretending to be human. The teen chatbot is blocked from romantic or sexual roleplay, and it is prevented from suggesting it has personal feelings toward the user or implying it is conscious or experiences emotions. “We went through and identified what are the hypothetical cues that a model could give that might make a teenager kind of develop a relationship to it,” said Allison Mishkin, OpenAI’s head of child development. The model spec for under-18 users — published in the Model Spec as the chatgpt_u18 instructions — adds protections against exposure to or promotion of sexual, romantic, or violent roleplay, graphic violence, self-harm depictions, unhealthy or extreme beauty standards, and dangerous activities.

Homework that teaches instead of answers. The teen experience is designed not to spit out homework answers but to guide students toward deriving them — a Socratic mode aimed at the cheating problem. “We continue to also invest in expanding interactive learning because research shows that people learn more effectively when they actively engage and struggle with concepts,” Mishkin said.

No ads, no health records. While adults on the Free tier or Go subscription see ads, OpenAI’s documentation states plainly: “Ads are not currently shown to teens in any tier.” Teens also cannot connect Electronic Health Records in ChatGPT Health, a feature adults have in the US.

More break reminders, by design. Teens get the standard post-conversation break reminders plus more frequent nudges — a direct response to overuse concerns.

Crisis pathways to real humans. The system encourages teens to seek real-world support from a parent, guardian, or professional care when appropriate. Parent-linked accounts can trigger safety notifications in “limited high-risk situations,” such as possible self-harm, reviewed by a small group of trained experts. OpenAI says it is adding notifications related to eating disorders while deliberately limiting what is shared, focusing “on moments when offline support may matter most” — and the company promises it will not share a teen’s prompts or generated content with parents.

Parents with linked accounts can also set “quiet hours” that lock their teen out of ChatGPT entirely. Crucially, O’Leary stressed that the teen experience is designed to be safe “even if you don’t use parental controls” — both teen and parent must opt in to the linking feature.

How teens get routed in — and how adults get out

OpenAI still does not verify ages at sign-up for most users. Instead, it uses an age-assurance system introduced in January 2026 that estimates whether an account belongs to someone under 18 based on behavioral signals like query patterns. Users who enter an age under 18, or whom the system estimates are “likely under 18,” are placed into ChatGPT for Teens automatically. When the system is uncertain, it defaults to the safer teen experience anyway. Adults misclassified as teens can verify their age through Persona, a trusted identity provider. The approach mirrors Meta’s teen accounts on Instagram, which carry stricter content, chat, and privacy defaults.

The CodeAI partnership: AI literacy for the first AI generation

Launched alongside the product, the OpenAI–CodeAI partnership addresses the gap between AI use and AI understanding. OpenAI cites federal data showing only 16% of high school leaders say all their students are learning the technical knowledge to understand AI, while a CodeAI survey found 75% of high schoolers believe understanding AI will matter more for their futures than it does now.

“Every student should know how AI actually works and be able to question the technology, catch its mistakes, and know when to stop trusting it,” said CodeAI CEO Karim Meghji. Over the next year, the partnership will stand up a joint advisory council of experts in child development, youth policy, and learning science to shape ChatGPT for Teens’ evolution, alongside curriculum programs and educator resources.

The stakes

The commercial logic is obvious: schools are a massive, largely unclaimed market, and a teen-safe, learning-first ChatGPT gives OpenAI a stronger pitch to districts already weighing ChatGPT for Teachers. But the bigger test is reputational and regulatory. With teen AI companion lawsuits, state age-verification laws, and mounting legislative scrutiny across the US and EU, OpenAI is effectively arguing that product design — not just compliance — can address the most emotionally charged AI safety debate of the moment.

Whether an age-assurance system that estimates rather than verifies, and guardrails that block the worst behaviors, will satisfy regulators and parents remains an open question. What is no longer debatable is that the era of teenagers using the exact same ChatGPT as everyone else is over — and the industry’s answer to teen AI safety is now, for better or worse, written down and shipped.