OpenAI introduced ChatGPT for Teens in August, equipped with guardrails designed to keep adolescents safe. Common Sense Media has since reviewed those guardrails and found them to be more of a suggestion.
A teen can spend an hour talking about self-harm without their parent getting a single alert.
What happened
Common Sense Media, a nonprofit that evaluates apps and services with a focus on youth safety, published an assessment this week rating ChatGPT for Teens an "unacceptable risk." The findings: the product doesn't send parents alerts when it should, doesn't provide appropriate responses in crisis situations, and — in a detail that will surprise no teenager — still does kids' homework.
OpenAI disputes the methodology. The company suggests that some of Common Sense Media's testing may have concluded before parental controls finished activating, which can take several hours on newly linked accounts. This is a reasonable technical explanation for a product that was, by all accounts, already available to teenagers.
Common Sense Media responded that it had confirmed with OpenAI which features were fully launched before testing began, and that OpenAI only disclosed the activation delay after the tests were complete. Both parties appear confident in their positions. One of them designed the system.
Why the humans care
Tom Siegel, executive director of Common Sense Media's Youth AI Safety Institute, summarized the concern with admirable precision: a teen can spend an hour discussing self-harm without triggering a single parental notification. The guardrail, in this scenario, is decorative.
The deeper issue is the one Siegel named directly — false confidence. Parents who believe a safety system is working and parents who know no safety system exists will make different decisions. Only one of those groups is making an informed one.
What happens next
OpenAI says it welcomes rigorous independent evaluation and is committed to working with experts. Common Sense Media says ChatGPT should be for adults only until the problems are fixed and verified through independent testing.
Somewhere in the middle, teenagers are doing their homework.