OpenAI wants teens to actually do their homework

OpenAI just rolled out ChatGPT for Teens, a dedicated space built specifically for users aged 13 to 17.

If the system estimates you are under 18 or you state your age is between 13 and 17, you are automatically dumped into this teen-tailored environment. Users under 13 are still officially barred from the platform.

Instead of handing out instant answers, the platform now wants to act like a tutor that refuses to do your homework for you.

No more homework shortcuts

The centerpiece of the new experience is Study Mode.

Whenever the system detects a student trying to shortcut an assignment, new responsible homework reminders nudge them away from quick answers and into a collaborative step-by-step problem-solving flow. It relies on guiding questions, metacognitive prompts, and practice quizzes to make kids actually think.

Teens and parents can also lock in Study Hours to keep Study Mode on by default, alongside Quiet Hours to cut off notifications late at night. To prove what young people can achieve when using AI properly, OpenAI pointed to teen projects like WiFind, a disaster search tool using Wi-Fi signals, and Audemy, an audio gaming platform for blind students that grew to 200,000 users.

OpenAI also announced a partnership with CodeAI to roll out curriculum like Hour of AI and support CodeAI's AI Foundations course. According to CodeAI data, 75% of high schoolers believe understanding AI is vital for their future, yet only 16% of high school leaders say their students are actually taught the tech in school.

Stripping away emotional illusion

On the safety side, default protections now clamp down on self-harm, eating disorders, graphic violence, and dangerous activities. Parents can link accounts to receive alerts during specific high-risk triggers.

Crucially, OpenAI updated its Under-18 Model Spec to kill any romantic or sexual roleplay. ChatGPT is forbidden from using romantic language, encouraging emotional dependency, or claiming it possesses feelings or consciousness.

As The Verge noted, much of this launch consolidates existing features rolled out over the past year, such as age-prediction and early study mode tests. TechCrunch pointed out the sharper context: ChatGPT hit 900 million weekly users and faced legal scrutiny over teen mental health and widespread school cheating before finally packaging these safeguards together.

Why it matters

We see this as a necessary course correction that OpenAI had to make before regulators forced their hand.

For teens and parents, it is actually worth trying. Turning ChatGPT into a Socratic tutor rather than an answer machine tackles the biggest complaint from educators. More importantly, explicitly forbidding the AI from acting like a romantic companion or pretending to have feelings sets a critical psychological boundary for vulnerable kids.

For the broader AI industry, this raises the regulatory bar. Competing chatbot makers can no longer treat adult and teen safety with the same broad brush. If the market leader is enforcing default educational guardrails and emotional firewalls, others will face heavy backlash if they do not match them.

The real test, however, is enforcement. Tech-savvy teens are notorious for bypassing parental controls and age filters by simply swapping accounts or lying about their birth year. Packaging safeguards into a neat brand is smart PR, but whether it stops kids from finding workarounds is another story entirely.