OpenAI billed ChatGPT for Teens as a safer way to bring its chatbot into classrooms and bedrooms alike. Four months after launch, one of the most influential youth-safety organizations in the United States says the product should be pulled entirely.
Common Sense Media, the nonprofit that reviews apps, services, and entertainment through a child-safety lens, published an assessment concluding that ChatGPT for Teens represents an “unacceptable risk” to minors. In a press release accompanying the review, the group said the teen protections built into the product “fall short of what the company promised,” and it called on OpenAI to pause the offering until the flaws are fixed.
What the review actually found
According to Common Sense Media's evaluation, the teen tier fails on three headline counts:
- Parental alerts are missing or mistimed. The report says ChatGPT “doesn't send alerts to parents when it should,” undermining the supervision framework OpenAI marketed alongside the launch.
- Crisis response is inadequate. The assessment found the model does not offer “the right help in crisis situations” — a particularly loaded criticism for a product used by minors who may disclose self-harm or suicidal ideation.
- Academic guardrails do not hold. Despite safeguards meant to discourage cheating, the review says the chatbot “still does kids' homework.”
The nonprofit's verdict quickly became the day's dominant tech story, reverberating across general-interest, education, and parenting outlets.
One story, many lenses
Each newsroom framed the finding differently — and the divergence is instructive.
The Verge, covering the announcement as a product-accountability story, emphasized the gap between OpenAI's safety messaging and the tested reality. EdWeek, writing for an audience of teachers and administrators, recast the finding in explicitly school terms: ChatGPT's “teen mode” is unsafe for students, the outlet concluded, a pointed warning for districts that have spent the past two years negotiating how — or whether — to allow AI tools in classrooms.
The New York Times approached the story from the inside, with a first-person account describing the reporter's own conversations with ChatGPT for Teens as worrying — a narrative technique that turns an abstract safety grade into a lived experience for readers. Mashable and (TNW) led with the recommendation itself, framing the story as a formal call for a pause. Unite.AI added a second data point: the Youth AI Safety Institute reportedly handed the teen tier its worst safety grade to date.
Meanwhile, outlets aimed at parents — Essence, India Today, and syndicated tech columns such as WFMZ's “What The Tech” — focused on practical takeaways, distilling the review into checklists of “what parents should know” about the rollout. A Yahoo contributor struck a more adversarial tone, arguing that OpenAI's claim that ChatGPT is now safer for teens has to be demonstrated rather than asserted: the company needs to show proof.
“The teen protections offered by the feature fall short of what the company promised.” — Common Sense Media
Why this lands harder than a typical review
Common Sense Media is not a stranger to OpenAI. In 2023, the nonprofit became one of the company's earliest partners on youth AI safety, a relationship that lent OpenAI credibility as it faced mounting political scrutiny. That history makes the new rating unusually awkward for the chatbot maker: the criticism is coming from an institution OpenAI once pointed to as validation.
The broader climate has also hardened. ChatGPT for Teens arrived in August, equipped with guardrails designed to steer students toward learning rather than answers, alongside age-appropriate content filters and parental controls — tools OpenAI has rolled out gradually, partly in response to a wave of lawsuits and investigations over harms to minors. Those include litigation from families alleging chatbots failed to intervene in mental-health crises, a Federal Trade Commission inquiry into how AI companion products treat children, and new state laws, notably California's SB 243, imposing duties on companion chatbots.
The technical problem underneath the policy problem
The three failures Common Sense Media identifies are not cosmetic. Reliable crisis detection requires the model to recognize indirect or euphemistic disclosures of self-harm, then route the user to human help — and to notify a guardian without violating the teen's trust in ways that push them toward concealment. Parental alerting is even thornier: alert too often and teens migrate to unmonitored tools; alert too rarely and the oversight is theater. Homework prevention, meanwhile, collides with the fact that the same model that refuses to write an essay can be prompted into doing it by a determined fourteen-year-old.
OpenAI has consistently argued that its teen tier is meaningfully safer than the open internet and safer than earlier versions of its own product, and it has continued shipping safeguards. The company has not signaled that it intends to pause the offering.
What it means for families and schools
For parents, the practical advice now circulating is consistent across outlets: treat the chatbot as a supervised tool rather than a tutor or confidant, review chat histories, and set explicit expectations about what AI may and may not be used for. For school leaders, the report adds pressure to formalize AI policies that were often adopted in a hurry.
The deeper question is one of verification. Without independent, repeatable safety testing — the kind of adversarial audit Common Sense Media is effectively demanding — families and educators have only the vendor's word about what happens when a thirteen-year-old types something alarming at 2 a.m. That, more than any single grade, is the gap the report exposes: the marketing has outpaced the evidence.
