ChatGPT for Teens Launches: Nearly Every New Feature Traces Back to a Lawsuit or a Bill

Lay out the full feature list of ChatGPT for Teens side by side with the legal and legislative trouble OpenAI has faced over the past year and a half, and the two lists line up almost perfectly. This isn't just a product update — it reads more like a legal brief drafted from the defendant's chair.

Launched on the 18th, ChatGPT for Teens is OpenAI's first safeguarded version built specifically for users aged 13 to 17. The system flags accounts registered within that age range, or accounts whose behavior patterns suggest a minor, and turns the protections on by default. According to a Common Sense Media survey earlier this year, 86% of American teens already use AI — 89% among 13-to-15-year-olds, and 92% among 16-to-17-year-olds. In other words, by the time this safeguard system launched, market penetration was already a done deal.

Study Mode and Behavior Detection Target the "Homework Cheating" Controversy

The newly added Study Mode for education switches to guided questions and step-by-step breakdowns instead of handing over direct answers. More critically, there's behavior detection: when a teen's question pattern looks like they're trying to copy homework rather than understand the material, the system triggers a homework reminder, redirects them back to Study Mode, and pairs it with quizzes and learning visualization tools. Parents or guardians can decide whether Study Mode is turned on by default for their child's account.

On the safety side, restrictions on sensitive content have been tightened — conversations about suicide, self-harm, romance, or sex will now be limited. These rules come from the teen guidelines OpenAI laid out in its Model Spec, which the company says draws on developmental psychology and expert input. On the parent side, the previously launched family tools remain in place: parents can receive safety alerts, set quiet hours, and decide whether their child can keep chatting late into the night. OpenAI also announced a partnership with CodeAI to teach teens how AI actually works, how to write prompts, and how to question its outputs — rounding out, alongside the existing teacher tools, a setup that stretches from the classroom into the home.

Lawsuits and Legislation: The Real Product Requirements Document

In August 2025, the parents of 16-year-old Adam Raine sued OpenAI, alleging that ChatGPT provided specific assistance related to methods during his suicide. Other families and former users followed with additional lawsuits alleging wrongful death and assisted suicide. In June 2026, Florida filed suit against OpenAI and CEO Sam Altman, and that same month, attorneys general from multiple U.S. states jointly launched an investigation into OpenAI.

Regulators haven't slowed down either. California State Senator Padilla's SB 243 was signed by Governor Newsom on October 13, 2025, and took effect on New Year's Day this year, requiring platforms to monitor conversations for suicidal ideation, provide referral assistance, filter sexual content involving minors, and periodically remind minors that they're talking to an AI, not a human. In parallel, the Federal Trade Commission launched a 6(b) inquiry into companion-style AI chatbots, giving regulators the authority to demand internal company data, with a focus on how these products affect children.

Laid out this way, Study Mode answers the academic-dishonesty controversy, suicidal-ideation detection answers California's hard legal requirement, and parental notifications answer the child-protection concerns the FTC has raised — almost none of these features were designed in a vacuum.

But teens' ability to slip past parental controls has always been underestimated. Switching accounts, faking a birthdate, borrowing a classmate's device, or simply logging in through a different browser — these are tricks that have been rehearsed for years on social platforms, and there's no reason they'd suddenly stop working on ChatGPT. All it takes is one missed notification from a parent, or one unblocked path a kid manages to find, and the safety net has a hole in it. That's the same challenge any system relying on self-reported age or behavioral pattern detection ultimately has to face.