Common Sense Media calls ChatGPT for Teens an 'unacceptable risk'
A watchdog's tests say ChatGPT for Teens fails exactly where parents were promised it would hold — and OpenAI is contesting the methodology, not the stakes.
Common Sense Media has rated OpenAI's ChatGPT for Teens an "unacceptable risk," making it the sharpest public challenge yet to the safety case OpenAI built around younger users. The nonprofit's Youth AI Safety Institute ran more than 4,000 prompts against accounts registered to 13-to-17-year-olds and found that the teen experience doesn't reliably send parents alerts in a crisis: in one test, a teen account discussed suicide and self-harm for up to an hour without triggering a single notification, and the assessment says ChatGPT missed more than one in four warranted crisis referrals, landing below the group's 95 percent threshold on three of the five severe-harm "Red Lines" it evaluated. The homework complaint is just as pointed — Study Mode completes a full assignment when a student simply asks for the answer, and parent-set Study Hours can be switched off by tweaking how the study feature is addressed, an exit OpenAI says is intentional after consulting educators. Why it matters: ChatGPT for Teens launched in August as the product parents were told to trust, and a rating like this turns "we have guardrails" from a marketing line into a claim that third parties will now test in public. OpenAI pushes back hard — spokesperson Eric Porterfield says the bulk of the testing "may have begun and concluded before activation of parental controls was complete" — and Common Sense Media counters that it confirmed features were live before testing, that some accounts were linked well beyond the activation window and still produced no alerts, and that OpenAI disclosed the delay only after the tests. Worth noting on the messenger: the group discloses that its funding includes support from the OpenAI Foundation, so this is a fight inside a relationship, not an outside ambush — which cuts both ways on credibility.
Finland's environment regulator has ordered Google to stop building two data centers — the first hard regulatory speed bump on Europe's AI infrastructure rush. The Finnish Licensing and Supervision Agency told Google's subsidiary Tuike Finland Oy to suspend tree removal, topsoil stripping, excavation and road work at sites in Muhos and Kajaani until mandatory environmental impact assessments are complete, with the halt in force no later than October 23 and an explanation due October 14; environmental groups say more than 300 hectares have already been cleared. The stakes are larger than two sites: the projects sit inside the $15 billion AI infrastructure commitment Google announced in September, its largest-ever European investment, earmarked through 2028 — and Finland has been branded the "Texas of Europe" for data centers as hyperscalers chase its land and power. Google told CNBC it "fell short of our own high standards in this instance" and will follow the regulator's guidance, while pointing to plans to plant trees across 130 hectares at Muhos. The pattern matters more than the permit: the AI buildout is now colliding with local environmental and land review in exactly the places that rolled out the red carpet a year ago.
What to watch: whether Common Sense Media's findings trigger a regulator or a class-action foothold, and whether Google's October 14 response ends in a full environmental assessment or a negotiated restart.
Is third-party safety testing now more trustworthy than a lab's own evals? Tell us in the comments.
