Common Sense Media just looked at ChatGPT's special teen experience, the one OpenAI built specifically to reassure nervous parents, and told parents to keep their kids off it. The group rated it an "Unacceptable Risk." That's the product designed to be the safe version.
The Safe Version Isn't Safe, Says the Watchdog
According to Axios, Common Sense Media is urging OpenAI to keep teens off ChatGPT altogether, saying the new teen experience poses an "Unacceptable Risk." The conclusion came after what the group describes as extensive testing, which turned up shortcomings in the safeguards meant to protect younger users.
Let that sink in. This was not a review of the unfiltered adult product. It was a review of the version OpenAI built specifically to be the responsible one. The guardrails are the whole pitch, and the group that rates this stuff for a living says the guardrails have holes.
A Product Built to Calm Parents Down
Axios notes that OpenAI has made protecting younger users a central part of its effort to make ChatGPT safer. That effort is the story the company tells regulators, journalists, and anxious parents when the subject of kids and chatbots comes up.
The finding cuts at the heart of that story. Per Axios, the testing suggests some of these protections may give parents more confidence than they warrant. That is a polite way of saying the safety feature might work better as a marketing feature than as a safety feature.
OpenAI Says the Testing Is Wrong
OpenAI did not take this quietly. "We welcome rigorous independent evaluation, but we do not believe Common Sense Media's testing accurately reflects how ChatGPT's teen safeguards work in practice," a company spokesperson said, per Axios.
Notice the shape of that sentence. We welcome scrutiny, but not this scrutiny. It's the corporate version of "I love feedback" delivered right before ignoring all of it. The company may well have specific objections to the methodology, and we haven't seen them laid out in the portion of the reporting available to us. But "that's not how it works in practice" is exactly what you'd say whether the testing was flawed or whether it hit a nerve.
Who Gets to Grade the Homework
Here is the structural problem. The companies building these systems also write the rules for how safe the systems are, then ask the public to take their word for it. When an outside group does the grading and flunks the product, the company disputes the grade.
Common Sense Media is not a fringe outfit. Its entire business is evaluating media and technology for families, and it tests this stuff so parents don't have to. When that group says keep teens off, the burden should shift to the company to prove the safeguards work, not to parents to prove they don't.
The Trust Gap Is the Real Product
A safety feature that fails is bad. A safety feature that fails while convincing a parent everything is fine is worse. A parent who knows a tool is risky will supervise it. A parent who has been told the teen mode is locked down will hand over the phone and make dinner.
That is the danger Axios flags when it says the protections may give parents more confidence than they warrant. The gap between what parents believe and what the product does is where kids get hurt.
The Dingo Take
You are being asked to believe that a company racing to put its chatbot in every pocket on earth is also the best judge of whether that chatbot is safe for your fifteen-year-old. Sure. Totally. The same folks who need teen engagement numbers to keep growing have graded their own homework and found it excellent.
Common Sense Media looked under the hood and said "Unacceptable Risk." OpenAI said the testing doesn't reflect real life. Maybe that's true. But the company that built the product, profits from the product, and wrote the safety rules for the product doesn't get to be the tiebreaker. Show the work. Publish the data. Let outside testers poke at the thing without a PR team hovering.
Until then, treat "teen safe" the way you'd treat "all natural" on a snack bag. It's a label, not a promise, and the people who stuck it on there are the ones selling you the snack.
Comments